Methane hydrate stability and anthropogenic climate changegeosci.uchicago.edu/~archer/reprints/archer.2007.hydrate_rev.pdf · D. Archer: Methane hydrate stability and anthropogenic

Biogeosciences, 4, 521–544, 2007www.biogeosciences.net/4/521/2007/© Author(s) 2007. This work is licensedunder a Creative Commons License.

Biogeosciences

Methane hydrate stability and anthropogenic climate change

D. Archer

University of Chicago, Department of the Geophysical Sciences, USA

Received: 20 March 2007 – Published in Biogeosciences Discuss.: 3 April 2007Revised: 14 June 2007 – Accepted: 19 July 2007 – Published: 25 July 2007

Abstract. Methane frozen into hydrate makes up a largereservoir of potentially volatile carbon below the sea floorand associated with permafrost soils. This reservoir intu-itively seems precarious, because hydrate ice floats in water,and melts at Earth surface conditions. The hydrate reservoiris so large that if 10% of the methane were released to the at-mosphere within a few years, it would have an impact on theEarth’s radiation budget equivalent to a factor of 10 increasein atmospheric CO2.

Hydrates are releasing methane to the atmosphere today inresponse to anthropogenic warming, for example along theArctic coastline of Siberia. However most of the hydratesare located at depths in soils and ocean sediments where an-thropogenic warming and any possible methane release willtake place over time scales of millennia. Individual catas-trophic releases like landslides and pockmark explosions aretoo small to reach a sizable fraction of the hydrates. Thecarbon isotopic excursion at the end of the Paleocene hasbeen interpreted as the release of thousands of Gton C, pos-sibly from hydrates, but the time scale of the release appearsto have been thousands of years, chronic rather than catas-trophic.

The potential climate impact in the coming century fromhydrate methane release is speculative but could be com-parable to climate feedbacks from the terrestrial biosphereand from peat, significant but not catastrophic. On geologictimescales, it is conceivable that hydrates could release asmuch carbon to the atmosphere/ocean system as we do byfossil fuel combustion.

Correspondence to:D. Archer([email protected])

1 Methane in the carbon cycle

1.1 Sources of methane

1.1.1 Juvenile methane

Methane, CH4, is the most chemically reduced form of car-bon. In the atmosphere and in parts of the biosphere con-trolled by the atmosphere, oxidized forms of carbon, such asCO2, the carbonate ions in seawater, and CaCO3, are moststable. Methane is therefore a transient species in our at-mosphere; its concentration must be maintained by ongoingrelease. One source of methane to the atmosphere is the re-duced interior of the Earth, via volcanic gases and hydrother-mal vents. Reducing power can leak from the interior of theEarth in other forms, such as molecular hydrogen, which cre-ates methane from CO2. The other source of reduced carbonis from photosynthesis, harvesting energy from sunlight. Byfar the greatest portion of the methane on Earth today wasgenerated originally from photosynthesis, rather than juve-nile release from the Earth.

Photosynthesis does not produce methane directly, be-cause methane as a gas has little use in the biochemical ma-chinery. Most biomolecules utilize carbon in an intermediateoxidation state, such as carbohydrates made up of multiplesof the unit CH2O with zero oxidation state, or on the re-duced end of the spectrum lipids with an oxidation state near–2. Once produced, biomolecules can be post-processed intomethane by one of two general pathways. One is biologi-cal, mediated by bacteria at low temperatures, and the otheris abiological, occurring spontaneously at elevated tempera-tures.

1.1.2 Biogenic methane

Biogenic methane is a product of organic matter degradation.Microbial respiration tends to utilize the partner electron ac-ceptor that will maximize the energy yield from the organic

Published by Copernicus Publications on behalf of the European Geosciences Union.

522 D. Archer: Methane hydrate stability and anthropogenic climate change

215

220

225

230

235

240

245

250

255

0 200 400 600 800 1000

Atmospheric Concentration, ppm

CO2

CH4

PresentConcentrations

Figure 1

Fig. 1. Radiative impacts of atmospheric methane and CO2 con-centrations: the outgoing longwave radiation flux over midlatitudewinter conditions, from the Modtran model (Rothman, 1992), witha web interface athttp://geosci.uchicago.edu/∼archer/cgimodels/radiation.html. The sensitivities to methane and CO2 are funda-mentally similar, but because methane is present at lower concen-tration, the atmosphere is at a steeper part of the curve where a sin-gle molecule of methane would have approximately twenty timesthe radiative impact of a single molecule of CO2. The leveling offof this curve is due to saturation of absorption bands.

matter. In the presence of molecular oxygen, O2, oxic respi-ration is the most energetically lucrative, and this is the path-way that is followed. With the depletion of O2, respirationproceeds using electron acceptors in the order NO−

3 , Mn2+,Fe2+, then SO2−

4 . Of these, SO2−

4 has potentially the highestavailability, because seawater contains high concentrationsof SO2−

4 . Once the SO2−

4 is depleted, methanogenesis canbegin. Fresh water has less SO2−

4 than seawater, so methano-genesis begins diagenetically earlier in fresh water systems.Fresh versus saltwater pathways can be distinguished by theirisotopic signatures ofδ13C andδD in the methane (Whiticarand Faber, 1986; Sowers, 2006). In sulfate-depleted salt wa-ter, the dominant pathway is the reduction of CO2 by molec-ular hydrogen, H2. H2 is produced bacterially by fermen-tation of organic matter, and is ubiquitous in marine sedi-ments, implicated in many other diagenetic reactions suchas iron, manganese, and nitrate reduction (Hoehler et al.,1999). Carbon isotopic values range from−60 to−100‰,while δD is typically −175 to−225‰. In fresh waters, thedominant pathway appears to be by the splitting of acetateinto CO2+CH4. Acetate, CH3COO−, can be produced frommolecular hydrogen, H2, and CO2 (Hoehler et al., 1999). TheH2 is produced by fermentation of organic matter (Hoehleret al., 1998). The isotopic signature is−40 to−50 in δ13C,and−300 to−350‰ inδD. Ultimately, by conservation ofoxidation state, if the source of reducing power is organicmatter, then a maximum of 50% of the organic carbon can beconverted to methane (Martens et al., 1998), by the reaction

2CH2O → CO2 + H2O (1)

In most sediments, for example at the Blake Ridge, bio-genic methane production is inferred to take place hundredsof meters below the depth where SO2−

4 is depleted, as indi-cated by linear gradients in SO2−

4 and CH4 as they diffuse to-ward their mutual annihiliation at the methane-sulfate bound-ary (Egeberg and Barth, 1998). At other locations methano-genesis is inferred to be occurring throughout the sulfate-richzone, but methane only accumulates to high concentrationswhen sulfate is gone (D’Hondt et al., 2002, 2004). Biologi-cal activity has been inferred to take place as deep as 800 mbelow the sea floor (D’Hondt et al., 2002, 2004; Wellsburyet al., 2002; Wellsbury et al., 2002; D’Hondt et al., 2004).

1.1.3 Thermogenic methane

As temperatures increase to about 110◦C degrees(Milkov, 2005), methane is produced, abiologically,from photosynthetically-produced organic matter. Thisthermogenic methane is distinguished by carbon isotopicvalues of about−30‰ (Whiticar and Faber, 1986), incontrast with the much lighter values,−60 to −110‰ ofbiogenic methane. Thermogenic methane is often associatedwith petroleum, coal, and other forms of fossil carbon.Petroleum is converted to methane if the deposits have everbeen buried deeper than the “oil window” of 7–15 km depth(Deffeyes, 2001). Most of the hydrates in the ocean derivefrom biogenic methane, but the Gulf of Mexico (Milkov,2005) and the Siberian gas fields (Grace and Hart, 1986)are examples of hydrate systems dominated by thermogenicmethane.

Thermogenic methane is also accompanied by otherlow-molecular weight organic compounds such as ethane(Milkov, 2005). In addition to serving as a tracer for theorigin of the methane, these compounds affect the thermo-dynamics of hydrate formation. Pure methane forms TypeI structural hydrates, while the inclusion of a few percentof ethane or H2S favors Type II structure. Type II hydratesare stable to 5–10◦C warmer, or perhaps 100 m deeper in thegeothermal gradient in warmer temperatures (Sloan, 1998).

1.2 Radiative impacts of methane release

1.2.1 Atmospheric release

CO2 is the dominant anthropogenic greenhouse gas in theatmosphere, because the anthropogenic perturbation to theCO2 concentration is much larger than the anthropogenicchange in CH4. However, the higher concentration of CO2means that on a per-molecule basis, CO2 is a less potentgreenhouse gas than CH4. Figure 1 shows the direct radiativeimpact of changes in CO2 and CH4 concentrations. The mostsignificant practical distinction between the gases is that CO2is more concentrated in the atmosphere than is methane. Forthis reason, in the strongest absorption bands of CO2, mostof the outgoing longwave light from the ground as already

Biogeosciences, 4, 521–544, 2007 www.biogeosciences.net/4/521/2007/

http://geosci.uchicago.edu/~archer/cgimodels/radiation.html

http://geosci.uchicago.edu/~archer/cgimodels/radiation.html

D. Archer: Methane hydrate stability and anthropogenic climate change 523

absorbed. An increase in CO2 concentration tends to makethe absorption bands a bit wider, but they cannot get anymore intense. Methane is less concentrated than CO2, and itsabsorption bands less saturated, so a single molecule of ad-ditional methane has a larger impact on the radiation balancethan a molecule of CO2, by about a factor of about 40 (basedon the total anthropogenic concentration changes of each gasand their radiative forcings from IPCC (2007)). Methane hasan indirect radiative effect, as its oxidation in the stratosphereproduces water vapor (Hansen et al., 2005). The radiativeimpact of CH4 follows the concentration to roughly the 1/3power, while the CO2 impact follows the log of the concen-tration. To get an idea of the scale, we note that a doublingof methane from present-day concentration would be equiv-alent to 60 ppm increase in CO2 from present-day, and 10times present methane would be equivalent to about a dou-bling of CO2. A release of 500 Gton C as methane (about10% of the hydrate reservoir) to the atmosphere would havean equivalent radiative impact to a factor of 10 increase inatmospheric CO2.

Once methane is released to the oxic, sunlit atmosphere,it oxidizes to CO2 on a time scale of about a decade. Ulti-mately, the oxidizing power comes from O2, but the reactivecompound OH is a necessary intermediate, following the re-action

CH4 + OH → CH3 + H2O (2)

where CH3 produced is a reactive radical compound, quicklyreacting with water vapor and other gases to form ultimatelyCO2. OH is produced by photolysis, the absorption of lightenergy by the severing of a chemical bond. Ozone photolyzesin the troposphere and combines with water to yield OH, asdo H2O2 and NO2. In the absence of sunlight, such as inice cores, no OH is produced, and CH4 and O2 are able tocoexist with negligible reaction for hundreds of thousands ofyears.

The implication of the short lifetime of methane in the at-mosphere is that the concentration of methane at any giventime is determined by the rate of methane emission over thepast few decades. If emission is steady with time, then thesteady-state atmospheric concentration can be expressed as

Inventory [mol]=

Emission flux[mol/year] × Atmospheric lifetime[years]

(3)

One uncertainty in this equation is how strongly themethane lifetime may depend on the methane source flux.If the methane oxidation rate is limited by the supply rateof OH, then an increase in the methane source flux couldincrease the methane lifetime in the atmosphere. In atmo-spheric photochemical models, a doubling of the source fluxresults in more than a doubling of the concentration (Pavlovet al., 2000).

The concentration of OH, and hence the lifetime andsteady-state concentration of methane, could be affected byanthropogenic emissions of combustion products such as thenitrogen oxides NOx, hydrocarbons, and carbon monoxide.In another atmospheric chemistry model (Wang and Jacob,1998), the concentrations of several gases have undergoneorder-one changes, but the effects of these changes on theOH concentration appear to largely cancel each other out.

If the methane is released quickly, on a time scale that isshort compared to the atmospheric lifetime, the methane con-centration spikes upward, decaying back toward the steady-state concentration. We will refer to a fast release as a “catas-trophic” methane release, as opposed to a long-term ongoingor “chronic” release. If the record of methane concentra-tion recorded in an ice core is undersampled or smoothedby diffusion within the fern or heterogeneous bubble closuredepth, then the maximum concentration of the event may notbe recorded in ice cores (Thorpe et al., 1996). The currentinventory of methane in the atmosphere is about 3 Gton C.Therefore, the release of 10 Gton C would triple atmosphericmethane.

1.2.2 Radiative impact of methane oxidized in the ocean

Once the methane is oxidized to CO2, it still acts as a green-house gas, albeit with less intensity. CO2 equilibrates be-tween the atmosphere and the ocean on a time scale of hun-dreds of years (Archer, 2005). Depending on the magnitudeof the CO2 release, i.e. the pH perturbation of the ocean, theequilibrium between the atmosphere and the ocean leaves15–30% of the CO2 release remaining in the atmosphere.This partitioning will apply whether the methane is oxidizedin the atmosphere or in the oxic ocean. If the methane is ox-idized in the atmosphere, the initial condition has more CO2in the atmosphere than at equilibrium, and the excess CO2will invade the ocean. Methane oxidized in the ocean willincrease the inventory of CO2 in the ocean, leading to grad-ual degassing of 15–30% over the coming centuries. Afteratmosphere-ocean equilibration, the distribution of the CO2between the atmosphere and ocean will be the same regard-less of whether the source of the CO2 was in the atmosphereor the ocean.

Excess CO2 in the atmosphere is gradually neutralized bydissolution of carbonate and silicate rocks, on time scalesthat range as long as 400 kyr (Archer, 2005). So, whilemethane is a transient species in the atmosphere, CO2 ac-cumulates. For this reason, the impact of a slow, ongoingmethane release might be to have greater radiative forcingfrom the accumulated CO2 than from the increased methaneconcentration, even while the methane release is ongoing(Fig. 2 from Archer and Buffett (2005), see also Harvey andHuang (1995), Schmidt and Shindell (2003)).

There exists an alternate pathway for methane oxidationthat does not produce CO2, but rather bicarbonate ion HCO−3 .

www.biogeosciences.net/4/521/2007/ Biogeosciences, 4, 521–544, 2007


0

0.5

1

1.5

2

0

1

2

3

4

0

2

4

6

0 20 40 60 80 1000

2

4

6

8

Time, kyr Time, kyr

1

10 100

1

10 100

1

10 100

1

10 100

CO2 Anthro + Clathrate

no ClathrateCH4

CO2 Anthro + Clathrate

no ClathrateCH4

300 Gton

1000 Gton

2000 Gton

5000 Gton

Figure 2

Fig. 2. Model projection of the radiative impact of a fossil fuel CO2release over the coming 100 kyr, from Archer and Buffett (2005).The model curves labeled 1, 10, and 100 assume methane releasetime constants from hydrates of 1, 10, and 100 kyr.(a) The warmingfrom the CO2 (300, 1000, 2000, and 5000 Gton releases) provokesmethane to degas from the ocean hydrate reservoir, increasing themethane concentration during the time interval that the methane isreleased. The methane is oxidized and accumulates in the atmo-sphere. (b) Radiative impacts of the CO2 and methane releases.The methane direct effect is smaller than the original CO2 radia-tive forcing, and it is also smaller and much shorter-lived than theradiative effect of the oxidized methane as CO2, gauged by the dif-ference between the anthropogenic CO2 radiative forcing with andwithout clathrate feedback. The point of the figure is to show thatthe greatest impact from a slow, ongoing methane release may befrom the accumulation of its oxidation product, CO2.

This is anaerobic oxidation of methane (AOM) (Boetius etal., 2000),

SO2−

4 + CH4 → HCO−

3 + HS−+ H2O (4)

The fate of the released alkalinity is often to precipitateas CaCO3. Authigenic CaCO3 has been used as a tracer forthe locations of CH4 emissions at Hydrate Ridge (Teichert etal., 2003, 2005), and invoked as an active player in the lifecycle of an emission field (Luff et al., 2005). A young ventsite should have an irregular, patchy distribution of carbon-ates, while an old site has become paved over with large flatCaCO3 slabs, which tend to seal off the methane emission(Sager et al., 1999).

2 The geology of methane hydrate

2.1 Methane production

The majority of the hydrate deposits on Earth are composedof biogenic methane, as indicated by its isotopic composi-

tion and the lack of other short hydrocarbons such as ethane.Most of the organic matter raining to the sea floor decom-poses in the top few centimeters of the sediment, calledthe zone of early diagenesis. However, the production ofmethane from this decaying organic matter is usually inhib-ited by the presence of dissolved sulfate, providing a moreenergetically favorable respiration pathway.

Sulfate is removed from pore waters deeper in the sed-iment by reaction with methane (anaerobic oxidation ofmethane, AOM, described above). This reaction preventssulfate and methane from coexisting at high concentrations insediment porewaters. Typically both species diffuse towardtheir mutual annihilation at a well-defined methane/sulfateboundary (Borowski et al., 1996, 1999; D’Hondt et al.,2004). After the depletion of sulfate, methane can be pro-duced from solid organic carbon, or by reaction of dissolvedorganic carbon, notably acetate, carried into the methano-genesis zone by diffusion or pore water advection. Wells-bury et al. (1997) found that heating sediment in the lab, upto 60◦C, stimulates the bacterial production of acetate. AtBlake Ridge, the concentration of acetate reaches very highconcentrations, supplying 10% of the reduced carbon neces-sary for methane production (Egeberg and Barth, 1998).

Bacterial abundances and metabolic rates of methanogen-esis, acetate formation, and AOM are extremely high at thebase of the hydrate and gas zone, rivaling metabolic rates atthe sediment surface (Parkes et al., 2000). Bacterial activ-ity is detected within the hydrate zone as well (Orcutt et al.,2004).

2.2 Methane transport

2.2.1 Diffusion

Once formed, methane moves within the sediment columnby diffusion, porewater flow, or migration of bubbles. Thetime scale for diffusion depends on the length scale as

T [s]=1x2[m2

]/D[m2/s] (5)

whereD is a diffusion coefficient, of order 10−9, 10−6, and10−4 m2/s for a solute, for heat, and for pressure, respec-tively. Heat can diffuse approximately 100 m in about 300years (Fig. 3). Solutes such as dissolved methane diffusemore slowly, while a pressure perturbation, such as wouldresult from decomposition of hydrate to yield methane bub-bles, diffuses away more quickly.

Diffusion is slow enough to insulate most of the hydratereservoir from anthropogenic warming in the coming cen-tury. Hydrate melting to yield dissolved methane in porewa-ters, such as proposed by Sultan et al. (2004), is unlikely tohave much impact on climate for this reason also.

2.2.2 Aqueous flow

Pore-water flow has the potential to determine the distribu-tion of hydrates within the sediment column. One source of



fluid flow is the compaction of sediment as it is buried andsubjected to increasing lithostatic pressure. The degree ofcompaction of sediment grains is a conserved diagnostic ofthe maximum lithostatic pressure they have endured (Flem-ings et al., 2003). Subducting sediments on active marginsexpel water more quickly than on passive margins. Forma-tion of hydrates at the surface and dissolution at depth ap-parently also acts as a source of upward-flowing pore fluid atHydrate Ridge (Suess et al., 1999).

Pore water flow can be focused by layers of high per-meability in sediments (Hovland et al., 1997; Aoki et al.,2000; Flemings et al., 2002). Lateral flow steered by sedi-ment permeability predicts expulsion of fluid near the baseof the continental slope off New Jersey, consistent with theobserved patterns of porewater seeps, and leading to nucle-ation of landslides from the base of the slope, consistent withthe observation of submarine canyons on continental margins(Dugan and Flemings, 2000). Fluid flow of methane-bearingporewater might be regulated by the formation of authigeniccarbonate, blocking and steering the channels of flow (Luffet al., 2005).

Focused ongoing fluid effluent from the sediment intothe ocean generates a structure known as a mud volcano.Approximately 1800 mud volcanoes have been discoveredaround the world, above and below sea level, mostly in a beltbeginning in the Mediterranean Sea and winding across Asiato Indonesia (Dimitrov, 2002). Submarine mud volcanoesare often associated with methane hydrates (see below).

2.2.3 Gas migration

In addition to pore water flow, methane is able to migratethrough the sediment column as a gas. In cohesive sedi-ments, bubbles expand by fracturing the sediment matrix, re-sulting in elongated shapes (Boudreau et al., 2005). Bubblestend to rise because they are less dense than the water theyare surrounded by, even at the 200+ atmosphere pressures insediments of the deep sea. If the pressure in the gas phaseexceeds the lithostatic pressure in the sediment, fracture andgas escape can occur (Flemings et al., 2003). Modeled andmeasured (Dickens et al., 1997) pressures in the sedimentcolumn at Blake Ridge indicate that this may be going on.

There is a differential-pressure mechanism which beginsto operate when the bubbles occupy more than about 10%of the volume of the pore spaces (Hornbach et al., 2004). Ifa connected bubble spans a large enough depth range, thepressure of the pore water will be higher at the bottom of thebubble than it is at the top, because of the weight of the porewater over that depth span. The pressure inside the bubblewill be more nearly constant over the depth span, becausethe compressed gas is not as dense as the pore water is. Thiswill result in a pressure gradient at the top and the bottom ofthe bubble, tending to push the bubble upward. Hornbach etal. (2004) postulated that this mechanism might be respon-sible for allowing methane to escape from the sediment col-

10

100

10001 102 10810610410-2

Ventilation Time Scale, years

Solutes

Heat

Pressure

Figure 3

Fig. 3. The diffusive time scale, given by (Dt)1/2, whereD is adiffusion coefficient andt is time, plotted as a function of distance,for heat, pressure, and solutes.

umn, and calculated the maximum thickness of an intercon-nected bubble zone, before the bubbles would break throughthe overlying sediment column. In their calculations, and instratigraphic deposits (they refer to them as “basin settings”)the thickness of the bubble column increases as the stabil-ity zone gets thicker. It takes more pressure force to breakthrough a thicker stability zone, so a taller column of gasis required. In compressional settings, where the dominantforce is directed sideways by tectonics, rather than down-ward by gravity, the bubble layer is never as thick, reflectingan easier path to methane escape.

There are multiple lines of evidence that gas can be trans-ported through the hydrate stability zone, without freezinginto hydrate. Seismic studies at Blake Ridge have observedthe presence of bubbles along faults in the sediment matrix(Taylor et al., 2000). Faults have been correlated with sites ofmethane gas emission from the sea floor (Aoki et al., 2000;Zuhlsdorff et al., 2000; Zuhlsdorff and Spiess, 2004). Seis-mic studies often show “wipeout zones” where the bubblezone beneath the hydrate stability zone is missing, and all ofthe layered structure of the sediment column within the sta-bility zone is smoothed out. These are interpreted to be areaswhere gas has broken through the structure of the sedimentto escape to the ocean (Riedel et al., 2002; Wood et al., 2002;Hill et al., 2004). Bubbles associated with seismic wipeoutzones are observed within the depth range which should bewithin the hydrate stability zone, assuming that the tempera-ture of the sediment column is the steady-state expression ofthe local average geothermal gradient (Gorman et al., 2002).This observation has been explained by assuming that up-ward migration of the fluid carries with it heat, maintain-ing a warm channel where gas can be transported throughwhat would otherwise be thermodynamically hostile territory



Depth(km

)

Depth(km

)Depth

(km)

Depth(km

)

Figure4

Fig. 4. The methane hydrate stability zone in surface sediments.Hydrate solubility temperature is given by the long-dashed line.Offshore water column temperatures are from Levitus et al. (1993),given by the solid black lines. Nearshore, the sea floor impinges onthe water column, so that temperature follows the geotherm (shortdashes). The thickness of the stability zone (heavy solid lines) in-creases with ocean depth.

(Taylor et al., 2000; Wood et al., 2002).The potential for gas migration through the stability zone

is one of the more significant uncertainties in forecasting theocean hydrate response to anthropogenic warming (Harveyand Huang, 1995).

2.3 Physical chemistry of methane hydrate

2.3.1 Thermodynamics

If the dissolved methane concentration reaches the saturationvalue for hydrate formation at the local temperature and pres-sure conditions, methane and water will freeze together intomethane hydrate or clathrate deposits. Thermodynamically,the stability of hydrate is determined by the temperature andby the availability of methane. The phase boundary is usuallydrawn assuming the presence of bubbles of pure methane, sothat the partial pressure of methane is determined by the to-tal fluid pressure. The partial pressure of methane dissolvedin oxic seawater is vanishingly small, but if hydrate wouldbe stable given the presence of methane bubbles, we callthat the phase boundary of hydrate stability in Fig. 4. Atatmospheric pressure, hydrate is never stable at Earth surfacetemperatures. At water depths of 100 m, hydrate would format about –20◦C, while at 500 m depth, the melting tempera-ture approaches in-situ temperatures. This minimum stabil-ity depth is somewhat shallower in the high-latitude oceans,about 200 m in the Arctic Ocean, because the upper water

column is colder (Fig. 4). In some locations, such as underthe sealed-off ice complex in Siberia, or in rapidly deposit-ing or low permeability sediments, the fluid pressure can beinfluenced by the weight of solids, and the fluid pressure willapproach the lithostatic pressure rather than the hydrostaticpressure. The stability depth for hydrate in permafrost inthe lithostatic case is about 200 m (Buffett, 2000), but hy-drate has been inferred to exist shallower than that, sealedinto “ice-bonded” permafrost (Dallimore and Collett, 1995).

2.3.2 Kinetics

Hydrate can persist metastably, several degrees above itsthermodynamic melting temperature, because of the energybarrier of nucleating small bubbles of methane gas (Buffettand Zatsepina, 1999). Rapid depressurization such as occursduring core retrieval does lead to melting of hydrate (Cir-cone et al., 2000). The dissolution of hydrates appears to bediffusion-controlled (Rehder et al., 2004). In general, kineticeffects are probably of secondary importance for predict-ing the hydrate response to anthropogenic climate change,because the thermal forcing takes place on such long timescales.

Lab experiments show that hydrate can nucleate from thepure aqueous phase, with no bubbles required, helping thecreation of hydrate from advective or biogenic methane (Buf-fett and Zatsepina, 2000). Several studies (Clennell et al.,1999, 2000; Lorenson, 2000) predict inhibition of hydrateformation in fine-grained sediment caused by the high ac-tivation energy of forming small crystals in the hydropho-bic small cavities of the pore water. This would explain thecharacteristic textures of hydrate: as pore-filling cement incoarse-grained sediment, but as irregularly shaped masses ofpure hydrate in fine-grained sediment, and predicts that hy-drates should form first or predominantly in sandy sediments(Lorenson, 2000; Winters et al., 2004).

2.4 Mechanisms of methane release

2.4.1 Deep ocean temperature change

The time-dependence of changes in the inventory of methanein the hydrate reservoir depends on the time scales of tem-perature and chemical processes acting. Figure 5 shows theapproximate time scales for altering the temperatures of theocean, as a function of depth. There is evidence from pa-leotracers (Martin et al., 2005) and from modeling (Archeret al., 2004) that the temperature of the deep sea is sensitiveto the climate of the Earth’s surface. Some warming has al-ready been detected in intermediate depths of the ocean inresponse to anthropogenic greenhouse gas warming (Levi-tus et al., 2003). In general, the time scale for changing thetemperature of the ocean increases with depth. There aresignificant regional variations in the ventilation time of theocean, and in the amount of warming that might be expected



in the future. As has already been mentioned, the Arctic isexpected to warm particularly strongly, because of the albedofeedback from melting the Artic ice cap. Temperatures in theNorth Atlantic appear to be sensitive to changes in ocean cir-culation such as during Dansgaard Oeschger climate events(Dansgaard et al., 1989).

As described above, the top of the hydrate stability zone isat 200–600 m water depth, depending on the temperature ofthe water column and the solubility of the hydrate under thelocal chemical conditions (sulfide, hydrocarbons). Withinthe sediment column, the temperature increases with depthalong the geothermal temperature gradient, 30–50◦C/km(Harvey and Huang, 1995). The shallowest sediments thatcould contain hydrate only have a thin hydrate stability zone,and the stability zone thickness increases with water depth.

A change in the temperature of the deep ocean will actas a change in the upper boundary condition of the sedi-ment temperature profile. Warming of the overlying oceandoes not put surface sediments into undersaturation, but thewarmer overlying temperature propagates downward until anew profile with the same geothermal temperature gradientcan be established. How long this takes is a strong (secondorder) function of the thickness of the stability zone, but thetime scales are in general long. In 1000 years the temperaturesignal should have propagated about 180 m in the sediment(Fig. 3). In the steady state, an increase in ocean tempera-ture will decrease the thickness of the stability zone. Dickenset al. (2001) calculated that the volume of the stability zoneought to decrease by about a factor of 2 with a temperatureincrease of 5◦C.

After an increase in temperature of the overlying watercauses hydrate to melt at the base of the stability zone, thefate of the released methane is difficult to predict. The in-crease in pore volume and pressure could provoke gas migra-tion through the stability zone (see Section 2.2.3) or a land-slide, or the bubbles could remain enmeshed in the sedimentmatrix. Hydrate is carried to the base of the stability zoneby the accumulation of sediment at the sea floor, so meltingof hydrate at the stability zone takes place continuously, notjust associated with ocean warming.

2.4.2 Pockmarks

The sediment surface of the world’s ocean has holes init called pockmarks (Hovland and Judd, 1988; Hill et al.,2004), interpreted to be the result of catastrophic or con-tinuous escape of gas to the ocean. Pockmarks off Norwayare accompanied by authigenic carbonate deposits associatedwith anerobic oxidation of methane (Hovland et al., 2005).Pockmarks range in size from meters to kilometers (Hovlandet al., 2005), with one 700 km2 example on the Blake Ridge(Kvenvolden, 1999). If the Blake Ridge pockmark is the re-sult of a catastrophic explosion, it could not have releasedmore than a Gton C as methane (assuming a 500 m thicklayer of 4% methane yields 1 Gton). Pockmark methane

0

1

2

3

4

50.1 1 10 100 1000

SurfaceThermocline

Intermediate

Deep

Ventilation Time Scale, years

a

Figure 5

Fig. 5. A rough estimate of the ventilation time of the ocean as afunction of ocean depth. Shallow waters warm in response to cli-mate change more quickly than deep waters. Ventilation times ofthe real ocean vary laterally, as well; the North Atlantic, for exam-ple, ventilates more quickly than the ocean average because of thepathway of subsurface flow in the ocean.

emission is most significant as an ongoing “chronic” sourcerather than single “catastrophic” releases.

2.4.3 Landslides

Another mechanism for releasing methane from the sedimentcolumn is by submarine landslides. These are a normal, in-tegral part of the ocean sedimentary system (Hampton et al.,1996; Nisbet and Piper, 1998). Submarine landslides are es-pecially prevalent in river deltas, because of the high rate ofsediment delivery, and the presence of submarine canyons.The tendency for slope failure can be amplified if the sed-iment accumulates so quickly that the excess high poros-ity of surface sediments cannot be squeezed out. This canlead to instability of the sediment column, causing periodicStoregga-type landslides off the coast of Norway (see be-low), in the Mediterranean Sea (Rothwell et al., 2000) andpotentially off the East coast of the United States (Dugan andFlemings, 2000). Maslin et al. (2004) find that 70% of thelandslides in the North Atlantic over the last 45 kyr occurredwithin the time windows of the two meltwater peaks 15–13and 11–8 kyr ago. These could have been driven by deglacialsediment loading or warming of the water column triggeringhydrate melting.

Warming or sea level fall may trigger the melting ofhydrate deposits, provoking landslides (Kvenvolden, 1999;Driscoll et al., 2000; Vogt and Jung, 2002). Paul (1978)calculates that landslides can release up to about 5 Gton Cas methane, enough to alter the radiative forcing by about0.2 W/m2. The origin of these estimates is discussed in thesection on the Storegga Slide, below.



2.4.4 Fate of methane released as bubbles

Methane released from sediments in the ocean may reach theatmosphere directly, or it may dissolve in the ocean. Bub-bles are not generally a very efficient means of transport-ing methane through the ocean to the atmosphere. Rehderet al. (2002) compared the dissolution kinetics of methaneand argon, and found an enhanced lifetime of methane bub-bles below the saturation depth in the ocean, about 500 m,because a hydrate film on the surface of the methane bubblesinhibited gas exchange. Bubbles dissolve more slowly frompetroleum seeps, where oily films on the surface of the bub-ble inhibits gas exchange, also changing the shapes of thebubbles (Leifer and MacDonald, 2003). On a larger scale,however, Leifer et al. (2000) diagnosed that the rate of bub-ble dissolution is limited by turbulent transport of methane-rich water out of the bubble stream into the open water col-umn. The impact of the surface dissolution inhibition onmethane transport through the water column seems small;in the Rehder et al. (2002) study a 2 cm bubble dissolves in30 m above the stability zone, and only 110 m below the sta-bility zone. Acoustic imaging of the bubble plume from Hy-drate Ridge showed bubbles surviving from 600–700 m wa-ter depth where they were released to just above the stabilityzone at 400 m (Heeschen et al., 2003). One could imaginehydrate-film dissolution inhibition as a mechanism to con-centrate the release of methane into the upper water column,but not really as a mechanism to get methane through theocean directly to the atmosphere.

Methane can reach the atmosphere if the methane bubblesare released in waters that are only a few tens of meters deep,as in the case of melting ice complex permafrost off Siberia(Xu et al., 2001; Shakhova et al., 2005; Washburn et al.,2005) or during time periods of lower sea level (Luyendyk etal., 2005). If the rate of methane release is large enough, therising column of seawater in contact with the bubbles maysaturate with methane, or the bubbles can be larger, poten-tially increasing the escape efficiency to the atmosphere.

2.4.5 Fate of methane hydrate in the water column

Pure methane hydrate is buoyant in seawater, so floating hy-drate is another potential way to deliver methane from thesediment to the atmosphere (Brewer et al., 2002). In sandysediment, the hydrate tends to fill the existing pore structureof the sediment, potentially entraining sufficient sediment toprevent the hydrate/sediment mixture from floating, while infine-grained sediments, bubble and hydrate grow by fractur-ing the cohesion of the sediment, resulting in irregular blobsof bubbles (Gardiner et al., 2003; Boudreau et al., 2005) orpure hydrate. Brewer et al. (2002) and Paull et al. (2003) triedthe experiment of stirring surface sediments from HydrateRidge using the mechanical arm of a submersible remotelyoperated vehicle, and found that hydrate did manage to shedits sediment load enough to float. Hydrate pieces of 0.1 m

survived a 750 m ascent through the water column. Paull etal. (2003) described a scenario for a submarine landslide, inwhich the hydrates would gradually make their way free ofthe turbidity current comprised of the sediment / seawaterslurry.

2.4.6 Oxidation of dissolved methane in the ocean

Methane is unstable to bacterial oxidation in oxic seawater.Rehder et al. (1999) inferred an oxidation lifetime of methanein the high-latitude North Atlantic of 50 years. Clark etal. (2000) correlated methane emission from Coal Point inCalifornia with a methane maximum in the water column ex-tending into the Pacific Ocean. Methane oxidation is fasterin the deep ocean near a particular methane source where itsconcentration is higher (turnover time 1.5 years), than in thesurface ocean (turnover time of decades) (Valentine et al.,2001). Water-column concentration and isotopic measure-ments indicate complete water-column oxidation of the re-leased methane at Hydrate Ridge (Grant and Whiticar, 2002;Heeschen et al., 2005).

An oxidation lifetime of 50 years leaves plenty of time formethane gas to evaporate into the atmosphere. Typical gas-exchange timescales for gas evasion from the surface oceanwould be about 3–5 m per day. A surface mixed layer 100 mdeep would approach equilibrium (degas) in about a month.Even a 1000-m thick winter mixed layer would degas about30% during a three-month winter window. The ventilationtime of subsurface waters depends on the depth and the fluidtrajectories in the water (Luyten et al., 1983), but 50 yearsis enough time that a significant fraction of the methane dis-solving from bubbles might reach the atmosphere before it isoxidized.

2.5 Stratigraphic-type sedimentary hydrate deposits

The most common hydrate deposits on Earth are in theocean, and are the product of largely one-dimensional pro-cesses of organic carbon burial, bacterial methanogenesis,and methane transport in slow fluid flow. Following the ter-minology of Milkov and Sassen (2002), we will refer to theseas stratigraphic-type hydrate deposits.

In the steady state, the maximum concentration of hydrateis found at the base of the stability zone, with bubbles foundbelow (Davie and Buffett, 2001). Typical concentrations ofhydrate are a few percent of pore volume, and the amountof bubbles below the stability zone is also a few percent byvolume. The layer of bubbles is clearly apparent in seismicsections of the subsurface sediment. Temperature contourswithin the sediment column tend to parallel the sea floor,and so the layer of bubbles tends to parallel the sea floor aswell. For this reason, the bubble layer below the base of thestability zone is referred to as a “bottom simulating reflector”or BSR. Because it is remotely detectable, the distribution of



the BSR is one of the best indications of the distribution ofhydrates in sediments.

Most of the hydrate deposits on Earth correspond to thestratigraphic type, and hence the estimates of the global in-ventory of hydrates are based on the physics or on the ob-served distribution of these types of deposits. Estimatesrange from 500 to 10 000 Gton C as methane in hydrate glob-ally. The estimates can be compared according to two met-rics. One is the area of the sea floor where hydrates canbe found, and the other is the inventory of methane, as hy-drate and in some tabulations as bubbles, per square meter.Milkov (2004) does a detailed and very thorough compari-son of these characteristics of estimates, leaving no need formore than a summary of his results here.

The first piece of the estimate is the area of the sea floor,between 500 and 3000 m depth, that has high concentra-tion of hydrates. MacDonald (1990) assumed a high-hydrateslope-area fraction of 10%. Borowski et al. (1999) esti-mate that 30% of this area could contain hydrates, based onthe proportion of cores where porewater sulfate reaches zerowithin 50 m of the sea floor. Milkov (2004) views Borowski’s30% as an upper limit, and takes 10% as his best estimateof the high-hydrate slope area fraction. The Buffett andArcher (2004) model predicts nonzero column inventories ofmethane hydrate or bubbles in 16% of the area between 500–3000 m, but in much of that area the abundance of methaneis unmeasurably small. (Its presence is of interest thermo-dynamically, however.) If we take>0.25% over 200 m as adetectability cutoff, the model predicts 13% of the sea floorto fit that definition.

Many studies estimate the area containing hydrates basedon the organic carbon concentration of surface sediments.The critical cutoff organic carbon concentration is typi-cally taken as 1% (Kvenvolden, 1999) or 0.5% (Harveyand Huang, 1995), which correspond respectively to about15% or 30% of the area of sea floor between 500 and3000 m (based on results from the sediment diagenesis model(Archer et al., 2002) used in Buffett and Archer (2004)).Gornitz and Fung (1994) used high chlorophyll concentra-tions from Coastal Zone Color Scanner images as correlatesto the 0.5–1% organic carbon concentration in sediments.This assumption neglects the role of depth and oxygen indetermining the organic carbon degradation, but the satel-lite data generates a map that looks very similar to sedimen-tary organic carbon maps. The areas of the sea floor repre-sented by the Coastal Zone Color Scanner cutoffs were 13%and 32%, similar to the organic carbon areas for>1% and>0.5%. The surface organic carbon method is appealing be-cause of the general correspondence between surface organiccarbon and seismic indications of hydrates below, but thereare some caveats. The critical quantity in the Buffett andArcher (2004) model is the advective flux of organic carbonto the methanogenesis zone, which increases with increas-ing carbon concentration, but also depends on sedimentationrate, a boundary condition which must be accounted for in

some way. Using sediment surface organic carbon concen-trations neglects the possibility that conditions might havechanged in the millions of years it takes for surface sedi-ments to be advected to the methanogenesis zone (Fehn etal., 2000). Changes in ocean temperature could alter the crit-ical organic carbon concentration for hydrate formation inthe past (Buffett and Archer, 2004). In spite of these caveats,sediment organic carbon concentrations capture the generaltrend from oligotrophic to eutrophic, nearshore to pelagic, inthe ocean that also drive methane hydrate formation.

Another metric by which global methane inventory esti-mates can be compared is the volume fraction of methanehydrate within the porewater, averaged over the depth rangeof the hydrate stability zone. Kvenvolden (1988) assumed10%. Milkov (2004) argues for a value of 1.2%. The Buffettand Archer (2004) predicts about 1.5%. Data from the BlakeRidge range from 2–4% (Paull et al., 2000; Borowski, 2004).Values from Hydrate Ridge are lower, closer to 1%. Thecurrent data is probably too sparse to distinguish between1% and say 3% as a global average hydrate porewater vol-ume fraction, but the 10% volume fraction assumed in earlierstudies like the influential paper by Kvenvolden (1988) nowseems to be high, if Blake Ridge or Hydrate Ridge is takento be representative of the broader ocean.

There are two studies, Buffett and Archer (2004), andKlauda and Sandler (2005), based on mechanistic models ofthe sedimentary methane cycle. Both studies are based onthe 1-D column model of Davie and Buffett (2001). The twostudies differ in their global estimates by a factor of twenty.Klauda and Sandler (2005) estimate 76 000 Gton C in hy-drate, while Buffett and Archer (2004) predict 3000 Gton Cin hydrate (plus 2000 Gton C in bubbles). Both studies showa reasonable fit to data from the Blake Ridge. The differencecan be traced to differences in the sediment accumulationrate, and carbon conversion efficiencies, by the two studies.The Klauda and Sandler (2005) calculation assumes a uni-form accumulation rate of sediment, throughout the entireocean, of 10 cm/yr, too high for the deep sea by an order ofmagnitude and more. For this reason, the Klauda and Sandler(2005) model predicts that most of the hydrates in the oceanought to exist in abyssal sediments, rather than restricted tothe continental margins, as observed (in seismic studies, forexample).

Uncertainty in the areal coverage of methane hydrate sedi-ment contributes a factor of three uncertainty in our estimateof the global hydrate reservoir size. The average hydratefraction is also unknown to within a factor of three, result-ing in perhaps a factor of ten overall uncertainty. A potentialrange of hydrate inventories must span about 500–3000 GtonC, with the inclusion of bubble methane adding perhaps asimilar amount. The uncertainty range will be reduced in thefuture by (1) improvement in techniques for estimating theconcentration of methane, both as hydrate and as bubbles,ideally by seismic methods that provide regional coverage,and (2) by continued deep-core sampling within hydrate re-



gions, to constrain the variability in methane concentrations.

For comparison, the inventory of fossil fuels, mostly coal,is thought to be about 5000 Gton C (Rogner, 1997), compa-rable to the hydrate reservoir. The inventory of dissolvedoxidized carbon in the ocean (CO2, HCO−

3 , and CO=

3 ) isabout 38 000 Gton C. This sounds comfortably larger thanthe hydrate reservoir, but an addition of CO2 of this magni-tude on a fast time scale would be a sizable perturbation tothe pH of the ocean (Archer et al., 1997). The ocean containsabout 2×1017 moles of oxygen, which could be completelydepleted by reaction with about 1000 Gton C in methane.

2.6 Structural-type sedimentary hydrate deposits

In stratographic-type hydrate deposits, hydrate concentrationis highest near the base of the stability zone, often hundredsof meters below the sea floor. In shallower waters, where thestability zone is thinner, models predict smaller inventoriesof hydrate. Therefore, most of the hydrates in stratographic-type deposits tend to be deep. In contrast with this, ina few parts of the world, transport of presumably gaseousmethane, through faults or permeable channels, results in hy-drate deposits that are abundant at shallow depths in the sedi-ment column, closer to the sea floor. These “structural-type”deposits could be vulnerable to temperature-change drivenmelting on a faster time scale than the stratographic depositsare expected to be.

The Gulf of Mexico is a leaky oil field (Macdonald et al.,1994, 2002, 2004; MacDonald et al., 2002, 2004; Milkov andSassen, 2000, 2001, 2003; Sassen et al., 2001, 2003; Sassenand MacDonald, 1994). Natural oil seeps leave slicks on thesea surface that can be seen from space. Large chunks ofmethane hydrate were found on the sea floor in contact withseawater (Macdonald et al., 1994). One of the three chunksthey saw had vanished when they returned a year later; pre-sumably it had detached and floated away.

Collett (1998) estimate that 500 Gton C might reside as hy-drates in the Gulf sediments, but Milkov and Sassen (2001)estimate only 5 Gton C. In the Community Climate SystemModel model under doubled CO2 (after 80 years of 1%/yearCO2 increase, from C. Bitz, personal communication), wa-ters at 500 m depth in the Gulf warm about 0.75◦C, and 0.2◦

at 1000 m. In-situ temperatures at 500 m are much closer tothe melting temperature, so the relative change in the satura-tion state is much more significant at 500 m than deeper.

The equilibrium temperature change in the deep ocean toa large, 5000 Gton C fossil fuel release could be 3◦C (Archeret al., 2004). Milkov and Sassen (2003) subjected a 2-dimensional model of the hydrate deposits in the Gulf to a4◦C temperature increase and predicted that 2 Gton C fromhydrate would melt. However, there are no observations tosuggest that methane emission rates are currently accelerat-ing. Sassen et al. (2001) find no molecular fractionation ofgases in near-surface hydrate deposits that would be indica-

tive of partial dissolution, and suggests that the reservoir mayin fact be growing.

Other examples of structural deposits include the summitof Hydrate Ridge (Torres et al., 2004; Trehu et al., 2004)and the Niger Delta (Brooks et al., 2000). The distributionof hydrate at Hydrate Ridge indicates up-dip flow along sandlayers (Weinberger et al., 2005). Gas is forced into sandylayers where it accumulates until the gas pressure forces it tovent to the surface (Trehu et al., 2004). Trehu et al. (2004)estimate that 30–40% of pore space is occupied by hydrate,while gas fractions are 2–4%. Methane emerges to the seafloor with bubble vents and subsurface flows of 1 m/s, andin regions with bacterial mats and vesicomyid clams (Torreset al., 2002). Further examples of structural deposits includethe Peru Margin (Pecher et al., 2001) and Nankai Trough(Nouze et al., 2004).

2.6.1 Mud volcanoes

Mud volcanoes are produced by focused upward fluid flowinto the ocean. Mud volcanoes often trap methane in hydratedeposits that encircle the channels of fluid flow (Milkov,2000; Milkov et al., 2004). The fluid flow channels associ-ated with mud volcanoes are ringed with the seismic imagesof hydrate deposits, with authigenic carbonates indicative ofanoxic methane oxidation, and with pockmarks (Dimitrovand Woodside, 2003). Milkov (2000) estimates that mudvolcanoes contain at most 0.5 Gton C of methane in hydrate;about 100 times his estimate of the annual supply of methanefrom mud volcanoes.

2.7 Land deposits

The term permafrost is intended to distinguish whether wateris frozen, but it is defined in terms of temperature: a two-year mean annual temperature below 0◦C. It has been esti-mated that permafrost covers 20% of the terrestrial surfaceof the Earth. High-latitude northern permafrost is observedto be warming (Smith et al., 2005) and thawing (Payette etal., 2004; Camill, 2005). Ice near the surface can melt in thesummer, in what is called the “active zone”. Observationsshow that the active zone is getting thicker (Sazonova et al.,2004). When surface ice melts, soils collapse to form ter-rain called thermokarst (Nelson et al., 2002), and buildingsfall down (Stockstad, 2004). This process has had a dizzyingimpact on the subarctic landscape (Stockstad, 2004). Mod-els project 30–40% increase in active zone thickness by 2100(Stendel and Christensen, 2002), and a comparable decreasein the total area of permafrost soils (Anisimov and Nelson,1996). Melting is projected to be most intense in the marginalpermafrost zone in the south (Anisimov and Nelson, 1996)and along the Arctic ocean (Nelson et al., 2002).



2.7.1 Peat decomposition

Permafrost soils contain relict organic matter that surviveddecomposition due to the freezing temperatures. Fossil mam-moths, still edible after all these years, are examples of thisphenomenon (Herz, 1904). Peat deposits are a substantialreservoir of carbon, are estimated to be 350–450 Gton C(Stockstad, 2004). With a thaw will come accelerated de-composition of this organic matter, increasing the flux ofCO2 and CH4 (Liblik et al., 1997; Rivkina et al., 2000, 2004).Soil that has been frozen for thousands of years still containsviable populations of methanotrophic bacteria (Rivkina et al.,2004). The flux of methane from peat soils to the atmospherealso depends on the location of the water table, which con-trols the thickness of the oxic zone (Bubier et al., 1995, 2005;Liblik et al., 1997). If 20% of the peat reservoir convertedto methane, released over 100 years, this would release 0.7Gton C per year, doubling the atmospheric methane concen-tration.

2.7.2 Melting hydrates

There is also the possibility that methane is trapped inpermafrost-associated hydrate deposits, and might poten-tially be released upon melting of the permafrost. Permafrostsoil need not necessarily be continuous filled ice, it must sim-ply be below the freezing point of pure water. If the poresare open, in contact with the atmosphere, the pressure in thepore spaces will be hydrostatic with the fluid being the atmo-sphere. In this case, it will be virtually impossible to achievehigh enough pressures of methane to form hydrates underany reasonable temperature. However, if the pore space issealed, by ice or clay for example, then the lithostatic pres-sure will come to bear on the pore spaces. Minable naturalgas deposits are often at high pressure, demonstrating the im-pingement of the lithostatic pressure on the hydrostatic pres-sure. The higher lithostatic pressure is conducive to hydrateformation in places where methane is found.

The Messoyakh gas field, producing gas for 13 years inWestern Siberia, is thought to be mostly hydrate (Krason,2000). A profile of permafrost from the Mackenzie Deltashowed massive, visible hydrate at∼350 m, and implied thepresence of invisible pore-water hydrate crystals as shallowas about 120 m, in solidly ice-bonded sediment (Dallimoreand Collett, 1995). The stability zone is below∼250 m here.

Total amounts of hydrate methane in permafrost soilsare very poorly known, with estimates ranging from 7.5to 400 Gton C (estimates compiled by Gornitz and Fung(1994)).

There is a special case called the ice complex in Siberia(Romankevich, 1984; Hubberten and Romanovskii, 2001;Gavrilov et al., 2003). The ice complex is a sealed horizonof ice that was formed when sea level was as much as 120 mlower than today, during the last glacial maximum. Liquidground water froze as it flowed through the permafrost, cre-

Figure 6

Sea LevelStabilityZoneThickness

0

-20

-40

-60

-80

-100

-12020 15 10 5 0

0

40

80

120

160

200

Fig. 6. Modeling results replotted from Mienert et al (2005) of hy-drate stability in the vicinity of the Storegga slide off the coast ofNorway. Sea level rose, associated with the melting of the ice sheets(left axis). The stability zone thickness increased with the rise insea level, but then decreased because of warming (right axis). Thisparticular model scenario is for a temperature change from−1◦to4◦C, at 500 m. For all model scenarios, the landslide occurred sev-eral thousand years after hydrate destabilization by warming of thewater column.

ating a sealed layer up to 60–80 m thick, onshore and off-shore under as much as 100 m water depth. Bottom wa-ter temperatures are near freezing in these locations, and sothey currently do not provide much impetus to melt at thesurface, although surface melting may accelerate with fu-ture high-latitude warming. However, 0◦C is considerablywarmer than surface air temperatures during glacial times.A geotherm projected down from 0◦C intersects the melt-ing temperature at a much shallower depth than would ageotherm from a glacial surface temperature. For this rea-son, most of the melting of the submerged ice complex sincedeglaciation has been on the bottom of the ice, not on the top.

Melting is also driven by lateral invasion of the coastline,a melt-erosion process called thermokarst erosion (Gavrilovet al., 2003). The ice melts where it is exposed to the oceanalong the coast, collapsing the land into the sea and leavingmore ice exposed to melting. The Siberian coast has recededby 100–500 km in 7500 years (Hubberten and Romanovskii,2001), after the sea level finished its deglacial rise (see Fig. 6in Hubberten and Romanovskii (2001)). Entire islands havemelted within historical times (Romankevich, 1984).

Emission of hydrate-melt methane has been documentedalong the Siberian coastline. Coastal melting has resultedin 2500% supersaturation concentrations of methane relativeto the atmosphere in Siberian shelf waters (Shakhova et al.,2005). Two surveys of methane concentration, taken 1 yearapart, differed in their methane inventory by a factor of five.Whether this difference is due to differences in water cir-culation or methane degassing is unknown. Surface watersover the North Slope of Alaska were similarly supersaturated(Kvenvolden, 1999).



Table 1. Summary of mechanisms by which methane might escape to the atmosphere.

Chronic releases Inventory Potential Release Impacts

Stratigraphic-type hydratedeposits in the ocean

1000–6000 Any release would take millennia Effects would be most pronouncedon geological timescales. Radiativeeffect of accumulating CO2 >effectof transient methane.

Arctic Ocean Hundreds CH4 release ongoing today, but timescale for acceleration is probablydecades

Released to water column, couldreach atmosphere as CH4

Gulf of Mexico 5–500 Any release would take centuries Released to water column, small po-tential impact on atmospheric CH4

Peat Decomposition 350–450 Say 20% over 100 years Flux of 0.7 Gton/yr to the atmo-sphere, triples pCH4

Permafrost hydrate melting Hundreds Comparable to Peat Decomposition Comparable to Peat Decomposition

Catastrophic releases

Landslides 5 Gton from Storegga Some release as hydrate which canreach the atmosphere, but also bub-bles which dissolve in the water col-umn

The potential for methane release to the environment frompermafrost hydrate melting has not been extensively dis-cussed, but given the magnitude of the potential hydratereservoir, and the long time scale for melting, one couldimagine a chronic, ongoing release of methane that would ri-val the release of methane from decomposing peat (Table 1).

2.8 Hydrates as fossil fuel

Another pathway by which hydrate carbon might reach theatmosphere to affect climate is if it is combusted as a fos-sil fuel. Estimates of the total inventory of methane in hy-drate deposits globally are very high, but probably only asmall fraction of the hydrate reservoir would be extractable(Milkov and Sassen, 2002). The largest methane reservoir,the stratographic disseminated deposit, is the least attrac-tive economically. The concentration of methane is gener-ally too low for economical extraction using current tech-nology. The sediments of the Blake Ridge are impermeable(Kvenvolden, 1999), making extraction even more unlikely,while sediments in the Nankai Trough are more permeableand hence easier to extract (Milkov and Sassen, 2002), whichthe Japanese intend to do (Kerr, 2004). The other class ofoceanic deposits is the structurally-focused deposits, suchas found in the Gulf of Mexico (Milkov and Sassen, 2001)and mud volcanoes (Milkov, 2000). Milkov and Sassen(2001) estimate that the Gulf of Mexico contains 40 times asmuch hydrate methane as conventional subsurface reservoirmethane in that area.

The most likely near-term targets for methane hydrate ex-traction are deposits associated with permafrost soils on land

and in the shallow ocean. The Soviets drilled at least 50 wellsin the Messoyakha field, in which thermogenic methane istrapped under a dome of 450-m thick permafrost, one-thirdof it frozen into hydrates (Krason, 2000). The Soviets ex-tracted gas from Messoyakha for 13 years, injecting hot wa-ter and/or solutes (methanol and CaCl2) to destabilize hy-drate and release methane. Subsequently, an internationalconsortium lead by the Japanese drilled a series of wellson the north coast of Alaska, the Mallik field (Kerr, 2004;Chatti et al., 2005). Hydrates here are in a sandstone layer1000 m down, below mudstones. The hydrate-bearing sedi-ments were more permeable there than had been expected sothat methane could be extracted most economically by sim-ply reducing the pressure. Methane moved via fractures, andmore fractures could be broken with pressure spiking.

The prognosis for methane hydrate mining is that methanehydrates could supply perhaps 10% of global methane con-sumption in the coming decades, by analogy to coal-bedmethane 30 years ago (Grauls, 2001; Kerr, 2004). Methanehydrates could be a significant source of fossil energy, butnot limitless as might be inferred from the large estimates oftotal methane inventory in the global hydrate reservoir, sincemost of the hydrates are probably impractical to extract.

The possibility of geological hazard from methane drillinghas been discussed in a general way (Kvenvolden, 1999;Grauls, 2001; Chatti et al., 2005) but the likelihood ofmethane extraction causing slope instability still seems ratherspeculative. Some have considered replacing CH4 hydrateswith CO2 hydrates, sequestering CO2 and maintaining thestability of the continental slope in the process (Warzinskiand Holder, 1998). The Storegga slide (next section) was in-



vestigated in order to determine the safety of extracting gasfrom the Ormen Lange gas field within the Storegga slidearea.

3 Observations from the past

3.1 The Storegga landslide

3.1.1 Observations

One of the largest exposed submarine landslides in the oceanis the Storegga Slide in the Norwegian continental margin(Bryn et al., 2005; Mienert et al., 2000, 2005). The slide ex-cavated on average the top 250 m of sediment over a swathhundreds of kilometers wide, stretching half-way from Nor-way to Greenland. There have been comparable slides onthe Norwegian margin every approximately 100 kyr, roughlysynchronous with the glacial cycles (Solheim et al., 2005).The last one, Storegga proper, occurred about 8150 yearsago, after deglaciation. It generated a tsunami in what is nowthe United Kingdom (Smith et al., 2004). The Storegga slidearea contains methane hydrate deposits as indicated by a seis-mic BSR (Bunz and Mienert, 2004; Mienert et al., 2005;Zillmer et al., 2005a, b) corresponding to the base of the hy-drate stability zone at 200–300 m, and pockmarks (Hovlandet al., 2005) indicating gas expulsion from the sediment.

3.1.2 Inferences

The slide was presumably triggered by an earthquake, but thesediment column must have been destabilized by either orboth of two mechanisms. One is the rapid accumulation ofglacial sediment shed by the Fennoscandian ice sheet (Brynet al., 2005). As explained above, rapid sediment loadingtraps pore water in the sediment column faster than it canbe expelled by the increasing sediment load. At some point,the sediment column floats in its own porewater (Dugan andFlemings, 2000). This mechanism has the capacity to ex-plain why the Norwegian continental margin, of all places inthe world, should have landslides synchronous with climatechange.

The other possibility is the dissociation of methane hy-drate deposits by rising ocean temperatures. Rising sea levelis also a player in this story, but a smaller one. Rising sealevel tends to increase the thickness of the stability zone,by increasing the pressure. A model of the stability zoneshows this effect dominating deeper in the water column(Vogt and Jung, 2002); the stability zone is shown increas-ing by about 10 m for sediments in water depth below about750 m. Shallower sediments are more impacted by long-termtemperature changes, reconstructions of which show warm-ing of 5–6◦C over a thousand years or so, 11–12 kyr ago. Thelandslide occurred 2–3 kyr after the warming (Fig. 6 fromMienert et al. (2005)). The slide started at a few hundredmeters water depth, just off the continental slope, just where

Mienert et al. (2005) calculates the maximum change in thehydrate stability zone.

Sultan et al. (2004) claim that warming in the near-surfacesediment would provoke hydrate to dissolve by increasingthe saturation methane concentration. This form of disso-lution differs from heat-driven direct melting, however, inthat it produces dissolved methane, rather than methane bub-bles. Sultan et al. (2004) assert that melting to produce dis-solved methane increases the volume, although laboratorymeasurements of volume changes upon this form of meltingare equivocal, and in any case the volume changes are muchsmaller than for thermal melting that produces bubbles.

The amount of methane released by the slide can be esti-mated from the volume of the slide and the potential hydratecontent. Hydrate just outside the slide area has been esti-mated by seismic methods to fill as much as 10% of the porewater volume, in a layer about 50 m thick near the bottom ofthe stability zone (Bunz and Mienert, 2004). If these resultswere typical of the entire 104 km2 area of the slide, the slidecould have released 1–2 Gton C of methane in hydrate. Paul(1978) assumed 10% hydrate fraction and predicted 5 GtonC methane released. If 5 Gton C CH4 reached the atmo-sphere all at once, it would raise the atmospheric concentra-tion by about 2.5 ppm of methane, relative to a present-dayconcentration of about 1.7 ppm, trapping about 0.2 W/m2 ofgreenhouse heat. The methane radiative forcing would sub-side over a time scale of a decade or so, as the pulse ofreleased methane is oxidized to CO2, and the atmosphericmethane concentration relaxes toward the long-term steadystate value. The radiative impact of the Storegga landslidewould be comparable in magnitude but opposite in sign tothe eruption of a large volcano, such as the Mt. Pinatuboeruption (−2 W/m2), but it would last for longer (10 yearsfor methane and 2 for a volcano).

It is tantalizing to a paleoclimatologist to wonder if therecould be any connection between the Storegga landslide andthe 8.2 kyr climate event (Alley and Agustsdottir, 2005),which is presumed to have been triggered by fresh waterrelease to the North Atlantic. However, ice cores recorda 0.75 ppm drop in methane concentration during the 8.2kevent, not a rise. The shutdown of convection in the NorthAtlantic would have, if anything, cooled the overlying wa-ters. Thus there appears to be no link between the 8.2k eventand Storegga in either causal direction; methane released bythe landslide did not alter the climate, nor did the 8.2k eventcause the slide.

3.1.3 Implications

The modeling results of Mienert et al. (2005) in Fig. 4 raisea clear possibility that warming and melting of hydrates hadsomething to do with the slide. On the other hand, severalthousand years elapsed between the warming and the land-slide. This tends to argue against concern for such events inthe coming century. Estimates of potential methane emission



Figure 7

-1

0

+1

3

2

1

0

-1

-2

545556Age, million years

CO2 release

Deep oceanwarms

8

10

12

14

CO2recovers in~100,000 yrs

Ocean temp.recovers in~100,000 yrs

Fig. 7. Carbon and oxygen isotopic excursion known as the Pale-ocene Eocene Thermal Maximum, from (Zachos et al., 2001). Thedark line is an generalization of the raw data (light lines), from thesouth Atlantic and the western Pacific.

from a large landslide range from 1–5 Gton C, which is sig-nificant but not apocalyptic. The tsunami hazard might be acomparable cause for concern.

3.2 Paleocene Eocene thermal maximum

3.2.1 Observations

About 55 million years ago, theδ13C signature of carbonin the ocean and on land decreased by 2.5–5‰ on a timescale of less than 10 kyr, then recovered in parallel on a timescale of∼120–220 kyr (Kennett and Stott, 1991; Roehl etal., 2006; Farley and Eltgroth, 2003) (Fig. 7 from Zachos etal. (2001)). Theδ18O of CaCO3 from intermediate depthsin the ocean decreased by 2–3‰, indicative of a warmingof about 5◦C. The timing of the spikes is to a large extentsynchronous. Planktonic foraminifera and terrestrial carbonrecords show aδ13C perturbation a bit earlier than the ben-thics do, suggesting that the carbon spike invaded the deepocean from the atmosphere (Thomas et al., 2002). Similarevents, also associated with transient warmings although lesswell documented, have been described from other times ingeologic history (Hesselbo et al., 2000; Jenkyns, 2003).

3.2.2 Inferences

The benthicδ18O record is most easily interpreted as a tem-perature change, at a depth of several kilometers in the ocean,from about 8◦ to about 14◦C, in a few thousand years. Warm-ing is also implied by Mg/Ca ratios in CaCO3 (Zachos et al.,

2003). Planktonicδ18O can also be fractionated by changesin freshwater forcing, reflected in salinity, but salinity canbe assumed to be regionally more homogeneous in the deepocean than at the surface. There is usually an offset betweenthe δ18O recorded by foraminifera and the true equilibriumvalue, called a vital effect, but single-species reconstructionsof δ18O tend to reduce the impact of these vital effects. Theδ18O of the whole ocean changes when ice sheets grow, butthere were no ice sheets at this time.δ18O in chemical re-actions with rocks, but not on short time scales such as seenhere.

The change in carbon isotopic composition of the carbonin the ocean is attributed to the release of some amount ofisotopically light carbon to the atmosphere. However it is notclear where the carbon came from, or how much of it therewas. The magnitude of the carbon shift depends on whereit was recorded. The surface change recorded in CaCO3 insoils (Koch et al., 1992) and in some planktonic foraminifera(Thomas et al., 2002) is twice as large a change as is reportedfor the deep sea. Land records may be affected by changesin plant fractionation, driven by changing hydrological cy-cle (Bowen et al., 2004). Ocean records may be affected byCaCO3 dissolution (Zachos et al., 2005), resulting in diage-netic imprints on the remaining CaCO3, a necessity to usemultiple species, or simple inability to find CaCO3 at all.

We can estimate the change in the carbon inventory ofthe ocean by specifying an atmospheric pCO2 value, a meanocean temperature, and insisting on equilibrium with CaCO3(Zeebe and Westbroek, 2003). The ocean was warmer, priorto the PETM event, than it is today. Atmospheric pCO2was probably at least 560 ppm at this time (Huber et al.,2002). The present-day inventory of CO2 in the ocean isabout 40 000 Gton C. According to simple thermodynamics,neglecting changes in the biological pump or circulation ofthe ocean, the geological steady-state inventory for late Pa-leocene, pre-PETM time could have been on the order of50 000 Gton C.

The lighter the isotopic value, the smaller the amount ofcarbon that must be released to explain the isotopic shift(Fig. 8). Candidate sources include methane, which canrange in its isotopic composition from−30 to −110‰. Ifthe oceanδ13C value is taken at face value, and the sourcewas methane at−60‰, then 2000 Gton C would be requiredto explain the isotopic anomaly. If the source were thermo-genic methane or organic carbon atδ13C of about−25‰,then 6 000 Gton C would be required.

Buffett and Archer (2004) find that the steady-state hy-drate reservoir size in the ocean is extremely sensitive to thetemperature of the deep sea. At the temperature of Paleocenetime but with everything else as in the present-day ocean,they predict less than a thousand Gton C of methane in steadystate. As the ocean temperature decreases, the stability zonegets thinner and covers less area. Their model was able to fit6000 Gton C in the Arctic Ocean, however, using 6◦ temper-atures from CCSM (Huber et al., 2002) (which may be too



cold) and assuming that the basin had been anoxic (Sluijs etal., 2006).

Invoking the Arctic here solves a number of problems. Itis easier to imagine a large temperature change in an iso-lated, polar basin than it is in the whole ocean. This makesit easier to get a large synchronous release such as observed.Also, if methane is released in the Arctic, even if it oxidizedin the Arctic water column, in other parts of the ocean theδ13C signature would be found in planktonic forams beforebenthic, as observed in site 669 near the Antarctic. Bubblesreleased from the sea floor mostly dissolve before reachingthe atmosphere, and half the methane reservoir is bubbles(Buffett and Archer, 2004), so it is not easy to imagine get-ting methane from the main ocean sediments to the planktonbefore it reaches the deep sea.

Marine organic matter has an isotopic composition of−20‰, and would require 6000 Gton to explain the isotopicanomaly. Svensen et al. (2004) proposed that lava intrusionsinto organic-rich sediments could have caused the isotopicshift. They cite evidence that the isotopic composition ofmethane produced from magma intrusion should be−35 to−50‰, requiring therefore 2500–3500 Gton C to explain theisotope anomaly in the deep ocean. If CO2 were also re-leased, from metamorphism of CaCO3, the average isotopiccomposition of the carbon spike would be lower, and themass of carbon greater. Storey et al. (2007) showed thatthe opening of the North Atlantic Ocean corresponds in timewith the PETM. However, volcanic activity continued forhundreds of thousands of years, leaving still unexplained thereason for the fast (<10 000 years) carbon isotope excursion.

Comets are not well constrained in their isotopic compo-sitions, but cometary dust tends to be about−45‰ (Kent etal., 2003). Kent et al. (2003) calculate that an 11 km cometcontaining 20–25% organic matter, a rather large icy tarball,could deliver 200 Gton C, enough to decrease theδ13C of theatmosphere and upper ocean by 0.4‰. It is unlikely that acomet could deliver thousands of Gton C however. An im-pact strike to a carbonate platform or an organic-rich sedi-ment of some sort could release carbon, but it would take avery large crater to release thousands of Gton C.

Volcanic carbon has an isotopic composition of−7‰, re-quiring a huge carbon release of 20 000 Gton C. Excess car-bon emissions have been attributed to superplume cycles inthe mantle and flood basalt volcanism (Larson, 1991) How-ever, these events tend to take millions of years to play out(Dickens et al., 1995). Bralower et al. (1997) and Schmitzet al. (2004) find evidence of increased volcanism during thePETM interval, but view the volcanism as rearranging oceancirculation, triggering methane release, rather than a majorprimary source of carbon itself, presumably because the po-tential volcanic carbon source is too slow.

Acidification of the ocean by invasion of CO2 drove ashoaling of the depth of CaCO3 preservation in the Atlantic(Zachos et al., 2005) although curiously not in the Pacific(Zachos et al., 2003). The magnitude of the CCD shift in

-70-60-50-40-30-20-1000

5000

10000

15000

20000

25000

δ13Csource

Present-dayOcean

Paleocene Ocean

Figure 8

0

10000

20000

30000

40000

50000

Ocean δ13C Shift Land δ13C Shift

Fig. 8. Size of the implied PETM carbon release as a function of itscarbon isotopic composition. The right-hand axis assumes theδ13Cspike of−2.5‰ as observed in most oceanδ13C records. The left-hand axis assumes a globalδ13C shift from the paleosol records,about twice that in the ocean (Koch et al., 1992). The mass-balancecalculation requires an estimate of the ocean inventory of carbon;we show results assuming the present-day ocean carbon chemistry,and that for a Paleocene ocean assuming equilibrium with 600 ppmatmospheric CO2 and with solid CaCO3.

the Atlantic, if taken as representative of the whole ocean,would be suggestive of a large carbon addition, on the orderof 5000 Gton C or more (Archer et al., 1997).

A large carbon release is also supported by the warminginferred from theδ18O spike. The temperature can be al-tered by both CH4 and CO2. Schmidt and Shindell (2003)calculated that the steady-state atmospheric CH4 concen-tration during the period of excess emission (ranging from500–20 000 years) would be enough to explain the tem-perature change. However, the atmospheric methane con-centration anomaly would decay away a few decades afterthe excess emission ceased. At this point the temperatureanomaly would die away, hence as soon as the carbon iso-topic composition stopped plunging negative, the oxygenisotopic composition should recover. The carbon isotopiccomposition should remain light for hundreds of thousandsof years (Kump and Arthur, 1999) until it reapproached asteady-state value. The record shows instead that the oxygenand carbon isotopic anomalies recovered in parallel. Thissuggests that CO2 is the more likely greenhouse warmerrather than CH4. It could be that the time scale for the pCO2to reach steady state might be different than the time scalefor the isotopes to equilibrate, analogous to the equilibra-tion of the surface ocean by gas exchange: isotopes takelonger. However, in the Kump and Arthur (1999) model re-



sults, pCO2 seems to take longer to equilibrate thanδ13C.The first-order result is that the CO2 andδ13C timescales aremuch more similar than the CH4 andδ13C time scales wouldbe.

A warming of 5◦C would require somewhere between oneand two doublings of the atmospheric CO2 concentration, ifthe climate sensitivity is in the range of 2.5–5◦C. Beginningfrom 600 ppm, we would increase the pCO2 of the atmo-sphere to somewhere in the range of 1200–2400 ppm. Theamount of carbon required to achieve this value for hundredsof thousands of years (after equilibration with the ocean andwith the CaCO3 cycle) would be of order 20 000 Gton C. Thiswould imply a mean isotopic composition of the spike ofmantle isotopic composition, not isotopically light methane.The amount of carbon required to explain the observedδ18Owould be higher if the initial atmospheric pCO2 were higherthan the assumed 600 ppm. The only way that a biogenicmethane source could explain the warming is if the climatesensitivity were much higher in the Paleocene than it seemsto be today, which seems unlikely because the ice albedofeedback amplifies the climate sensitivity today (Pagani etal., 2006).

The bottom line conclusion about the source of the car-bon isotopic excursion is that it is still not clear. There is noclear evidence in favor of a small, very isotopically depletedsource of carbon. Mechanistically, it is easier to explain asmall release than a large one, and this is why methane hasbeen a popular culprit for explaining theδ13C shift. Radiativearguments argue for a larger carbon emission, correspond-ing to a less fractionated source than pure biogenic methane.Thermogenic methane might do, such as the explosion of alarger Gulf of Mexico, if there were a thermogenic depositthat large. Or perhaps it was some combination of sources, aninitial less-fractionated source such as marine organic matteror a comet, followed by hydrate release.

3.2.3 Implications

The PETM is significant to the present-day because it is aclose analog to the potential fossil fuel carbon release if weburn all the coal reserves. There is about 5000 Gton of C incoal, while oil and traditional natural gas deposits are hun-dreds of Gton each (Rogner, 1997). The recovery timescalefrom the PETM (140 kyr) is comparable to the model pre-dictions, based on the mechanism of the silicate weatheringthermostat (400 kyr timescale (Berner et al., 1983)).

The magnitude of the warming presents something of aproblem. 5000 Gton of fossil fuel release will warm the deepocean by perhaps 2–4◦C, based on paleoclimate records andmodel results (Martin et al., in press). The warming duringthe PETM was 5◦C, and this was from an atmospheric CO2concentration higher than today, at least 600 ppm and per-haps higher, so that a further spike of only 2000 Gton (basedon methane isotopic composition) would have only a tiny ra-diative impact, not enough to warm the Earth by 5◦C. One

possibility is our estimates for the climate sensitivity is un-derestimated by a factor of 2 or more. However, one mighthave expected a decreased climate sensitivity for an ice-freeworld than for the ice-age climate of today.

Another possibility is that the PETM was driven by twosources of carbon, totaling maybe 10 000 Gton C. At most10% of this carbon could have had an initialδ13C of −60‰,if the rest were volcanic carbon at−7‰. The implicationwould be that the hydrate reservoir at that time did not am-plify the initial carbon release (analogous to our fossil fuelCO2) by more than 10%. However, there are no strong ideasfor where that other 9000 Gton C could have come from.

Perhaps the global averageδ13C shift was as large asrecorded in soils (Koch et al., 1992) and some planktonicforaminifera (Thomas et al., 2002), and perhaps it was ther-mogenic methane, so the hydrate release could have been8000 Gton C. In this case we can attribute all of the tem-perature change to the radiative effect of the released car-bon, mostly as the accumulated CO2. The Archer and Buf-fett (2005) model predicted a regime in model space wherethe hydrate reservoir would be unstable, periodically meltingdown. The time period between meltdowns was determinedby the time scale of methane accumulation in the reservoir.The critical parameters to the model are the time scale for amelting relaxation to the equilibrium size, and the fraction ofthe reservoir which melts at all. If most of the reservoir equi-librates quickly, then periodic meltdowns result. Tauntingly,there are several tiny “after shocks” of the PETM, all about2 million years apart, such as an ELMO event (Lourens etal., 2005). The trouble then is that the model, tuned to peri-odic meltdowns during the PETM, predicts that the hydratereservoir today, larger because the ocean is colder, shouldperiodically melt down even more severely today.

Could some external agent of warming, not CO2, havedriven temperatures up? Theδ13C could be showing usmethane release, but the temperature would be attributed tosomething else, something no one has thought of yet. Thedifficulty here would be that the decay of the temperaturespike follows so closely the decay of theδ13C spike. Thistends to steer us back to the path of CO2 as the proximateagent of temperature change.

At present, the PETM serves as a cautionary tale aboutthe long duration of a release of new CO2 to the atmosphere(Archer, 2005). However, our current understanding of theprocesses responsible for theδ13C spike is not strong enoughto provide any new constraint to the stability of the methanehydrate reservoir in the immediate future.

3.3 Santa Barbara Basin and the Clathrate Gun Hypothesis

Kennett et al. (2003) and Nisbet (2002) argue that methanefrom hydrates could be responsible for the initial deglacialmethane concentration rise in the Greenland ice core record.Kennett et al. (2000) found episodic negativeδ13C excur-sions in benthic foraminifera in the Santa Barbara Basin,



which they interpret as reflecting release of hydrate methaneduring warm climate intervals. Biomarkers for methanotro-phy are found in greater abundance, and indicate greater ratesof reaction, during warm intervals in the Santa Barbara Basin(Hinrichs et al., 2003) and in the Japanese coastal margin(Uchida et al., 2004). Cannariato and Stott (2004) howeverargued that these results could have arisen from contamina-tion or subsequent diagenetic overprints.

Because methane is a transient chemical species in theatmosphere, a sustained increase in methane concentrationwould have to be driven by a sustained increase in the chronicmethane release rate to the atmosphere. A single catastrophicrelease of methane might not even be visible in the ice corerecords. Detailed analysis of the methane record shows alag of a few hundred years between abrupt warmingsandmethane rise. This pattern is consistent with a wetland sourcefor methane, but difficult to generate by catastrophic releasefrom hydrates (Brook et al., 2000). The interhemisphericgradient of methane tells us that the deglacial increase in at-mospheric methane arose from high Northern latitudes (Dal-lenbach et al., 2000).

The definitive ruling comes from the isotopic composi-tion of the ice core methane. The isotopic ratio of D/H inice core methane indicates a freshwater source, rather than amarine source, apparently ruling out much of a role for ma-rine hydrate methane release (Sowers, 2006). The deglacialmethane rise could therefore be attributed to methanogenesisfrom thawing organic matter decomposition or from high-latitude wetlands. Regardless of the source of the methane,the climate forcing from the observed methane record is tooweak to argue for a dominant role for methane in the glacialcycles. The climate forcings from changing atmosphericCO2, and the albedo forcing from melting ice, had a muchstronger impact on the evolution of climate.

4 Risks for the future

4.1 Capacity for doomsday

There is so much methane as hydrates on Earth that it seemslike a perfect ingredient for a climate doomsday scenario.Hydrate is unstable at Earth surface conditions, both becauseof the low atmospheric methane concentration and becausemost of the Earth’s surface is warmer than the freezing pointof methane hydrate at one atmosphere pressure. The hy-drate reservoir contains thousands of Gton C of methane,enough that releasing a small fraction of the methane directlyto the atmosphere, within a time window that is short rela-tive to the atmospheric lifetime of methane, could increasethe methane concentration of the atmosphere by a factor of100 to 1000 over pre-anthropogenic values. Methane ab-sorbs infrared light between about 1250 and 1350 cm−1, afrequency range at which terrestrial radiation is less intensethan it is in the absorption band of the CO2 bending mode,

about 600–700 cm−1. A massive increase in methane con-centration therefore has a smaller impact on the radiative bal-ance of the Earth than would a comparable increase in CO2,but nevertheless the greenhouse forcing from the methane in-crease could be catastrophic, equivalent to increasing CO2by a factor of 10 or more. The methane hydrate reservoirtherefore has the potential to warm Earth’s climate to Eocenehothouse-type conditions, within just a few years. The poten-tial for planetary devastation posed by the methane hydratereservoir therefore seems comparable to the destructive po-tential from nuclear winter or from a bolide impact.

Fortunately, most of the hydrate reservoir seems insolatedfrom the climate of the Earth’s surface, so that any melt-ing response will take place on time scales of millennia orlonger. Various potential mechamisms for releasing methanein response to climate change, discussed in detail above, aresummarized in Table 1.

4.2 Permafrost deposits

The Siberian margin is one example of a place wheremethane hydrate is melting today, presumably at an accel-erated rate in response to anthropogenic warming. This is aspecial case, where subsurface hydrates are exposed to theocean by lateral erosion of coastline. The coastline is re-ceding at rates of tens of meters per year in parts of Siberiaand Alaska, but this is an ongoing process that began withthe sea level rise of the deglaciation (Hubberten and Ro-manovskii, 2001). The melting of hydrates in this regionreleases methane in an ongoing, chronic way, potentially in-creasing the steady-state methane concentration of the at-mosphere, along with other ongoing anthropogenic methanefluxes. No mechanism has been proposed whereby a signif-icant fraction of the Siberian permafrost hydrates could re-lease their methane catastrophically.

4.3 Structural deposits

The most vulnerable hydrate deposits in the ocean appearto be the structural type, in which methane gas flows in thesubsurface, along faults or channels, perhaps to accumulateto high concentrations in domes or underneath impermeablesedimentary layers. The structural deposits have two dis-tinguishing characteristics that may affect their potential formethane release. First, they produce “massive” methane hy-drates, displacing the sediment to generate large chunks ofhydrate, potentially filling tens of percent of the volume ofthe sediment (Trehu et al., 2004), as opposed to just a fewpercent as in the stratigraphic-type hydrate deposits. The sig-nificance of this is that a large chunk of hydrate is more likelyto survive an ascent to the sea surface, if it escapes the sedi-ment column as a result of a submarine landslide (Brewer etal., 2002; Paull et al., 2003) or simply by breaking off fromthe sediment surface (Macdonald et al., 1994). The otherimportant characteristic of structural hydrate deposits is that



the hydrate can be found at shallower depths in the sediment,in general, than is typical for the stratigraphic-type deposits.Methane hydrates are found at the sediment surface in theGulf of Mexico (Macdonald et al., 1994) and Hydrate Ridge.Hydrate deposits of these characteristics is also often associ-ated with mud volcanoes.

The proximity of structural hydrate deposits to the watersof the ocean allows them to be affected by anthropogenicwarming without waiting thousands of years for heat to dif-fuse into the sediment column (Fig. 3). However, these de-posits are still covered with hundreds of meters of ocean wa-ter. Throughout most of the ocean, the stability depth is oforder 500–700 m, shoaling to perhaps 200 m in the Arctic.Surface warming is expected to take order a century to reachthese depths. Presumably any melting response to this grad-ual warming would be gradual as well, slower than the atmo-spheric lifetime of methane and therefore by our definition achronic methane release rather than a catastrophic one.

4.4 Stratigraphic deposits

Most of the hydrate deposits on Earth are of the stratigraphic-type, which implies that the hydrate is (1) typically diluteand (2) generally located near the base of the stability zonein the sediment, which can be hundreds of meters below thesea floor. Warming of the ocean can propagate into the sedi-ment column, but this thick layer of thermal insulation guar-antees that most of the anthropogenic effect on temperaturewill take thousands of years. Melting of hydrates in responseto warming will tend to occur primarily below the stabilityzone, where bubbles of methane will be produced in a pro-cess analogous to the ongoing melting of hydrates to producethe bubble-layer (BSR).

The fate of gas bubbles released in subsurface sedimentis still very uncertain. The gas could remain in place, or itcould escape through the cold trap of the stability zone, or itcould destabilize the sediment column, provoking submarinelandslides. A landslide could release methane as hydrate,which may reach the atmosphere, and bubbles, which proba-bly would not. A landslide methane release would certainlybe abrupt, but it would not be climatically catastrophic be-cause the amount of methane in any given landslide couldonly be a tiny fraction of the global methane inventory. Aslide the size of the Storegga slide off of Norway could po-tentially release enough methane to affect climate compara-bly to a large volcanic eruption, although it would be a warm-ing rather than a cooling, and it would last a decade ratherthan a few years. As it happens, there was no increase in at-mospheric methane during the time interval of the Storeggalandslide itself, but rather a decrease associated with the 8.2kclimate event, a cooling event triggered by a sudden meltwa-ter release to the ocean.

4.5 Century-timescale response

On the timescale of the coming century, it appears that mostof the hydrate reservoir will be insulated from anthropogenicclimate change. The exceptions are hydrate in permafrostsoils, especially those coastal areas, and in shallow oceansediments where methane gas is focused by subsurface mi-gration. The most likely response of these deposits to an-thropogenic climate change is an increased background rateof chronic methane release, rather than an abrupt release.Methane gas in the atmosphere is a transient species, itsloss by oxidation continually replenished by ongoing release.An increase in the rate of methane emission to the atmo-sphere from melting hydrates would increase the steady-statemethane concentration of the atmosphere.

The potential rate of methane emission from hydrates ismore speculative than the rate from other methane sourcessuch as the decomposition of peat in thawing permafrostdeposits, or anthropogenic emission from agricultural, live-stock, and fossil fuel industries, but the potential rates appearto be comparable between these sources.

4.6 Geological-timescale response

On geologic time scales, the strongest climate impact ap-pears to be from CO2, the oxidation product of any releasedmethane. Methane is a transient species in the atmosphere,with a lifetime of about a decade. CO2 accumulates in theatmosphere/ocean/terrestrial biosphere carbon pool, and per-sists to affect climate for hundreds of thousands of years(Archer, 2005). If a pool of methane is released over atimescale of thousands of years, the climate impact from theaccumulating CO2 concentration may exceed that from thesteady-state increase in the methane concentration (Archerand Buffett, 2005), see also Harvey and Huang (1995) andSchmidt and Shindell (2003). After the emission stops,methane drops quickly to a lower steady state, while the CO2persists (Schmidt and Shindell, 2003).

If hydrates melt in the ocean, much of the methanewould probably be oxidized in the ocean rather than reach-ing the atmosphere directly as methane. This reduces thecentury-timescale climate impact of melting hydrate, but ontimescales of millennia and longer the climate impact is thesame regardless of where the methane is oxidized. Methaneoxidized to CO2 in the ocean will equilibrate with the at-mosphere within a few hundred years, resulting in the samepartitioning of the added CO2 between the atmosphere andthe ocean regardless of its origin.

Archer and Buffett (2005) find an amplifying positivefeedback among atmospheric CO2, the temperature of thedeep ocean, and the release of carbon from methane hydrateto more atmospheric CO2. They find that if the melting kinet-ics of the reservoir are assumed to be too fast, or the releasefraction too high, then the reservoir becomes unstable, melt-ing down spontaneously in ways that are not found in the re-



centδ13C record. The rate and extent to which methane car-bon can escape the sediment column in response to warmingis, as we have seen, very difficult to constrain at present. Itdepends on the stability of the sediment slope to sliding, andon the permeability of the sediment and the hydrate stabilityzone cold trap to bubble methane fluxes. They find that ina worst-case scenario, after thousands or hundreds of thou-sands of years, the methane hydrate reservoir could releaseas much carbon as fossil fuel emissions.

Acknowledgements.I appreciate helpful reviews by E. Nisbet,B. Buffett, E. Thomas, A. Trehu, I. Fairchild, and two otheranonymous reviewers. This work was supported by the NationalScience Foundation and the German Advisory Council on GlobalChange (WBGU).

Edited by: K. Caldeira

References

Alley, R. B. and Agustsdottir, A. M.: The 8k event: cause and con-sequences of a major Holocene abrupt climate change, Quater-nary Sci. Rev., 24(10–11), 1123–1149, 2005.

Anisimov, O. A. and Nelson, F. E.: Permafrost distribution in theNorthern Hemisphere under scenarios of climate change, GlobalPlanet. Change, 14, 59–72, 1996.

Aoki, Y., Shimizu, S., Yamane, T., et al.: Methane hydrate accu-mulation along the western Nankai Trough, in: Gas Hydrates:Challenges for the Future, 136–145, 2000.

Archer, D.: Fate of fossil-fuel CO2 in geologic time, J. Geophys.Res. Oceans, C09S05, doi:10.1029/2004JC002625, 2005.

Archer, D., Kheshgi, H., and Maier-Reimer, E.: Multiple timescalesfor neutralization of fossil fuel CO2, Geophys. Res. Lett., 24,405–408, 1997.

Archer, D., Martin, P., and Lea, D.: The importance of ocean tem-perature to global biogeochemistry, Earth Planet. Sci. Lett., 222,333–348, 2004.

Archer, D. E. and Buffett, B.: Time-dependent response ofthe global ocean clathrate reservoir to climatic and anthro-pogenic forcing, Geochem. Geophys. Geosys., 6(3), Q03002,doi:10.1029/2004GC000854, 2005.

Archer, D. E., Morford, J. L., and Emerson S.: A model ofsuboxic sedimentary diagenesis suitable for automatic tuningand gridded global domains, Global Biogeochem. Cy., 16,doi:10.1029/2000GB001288, 2002.

Berner, R. A., Lasaga, A. C., and Garrels, R.M.: The carbonate-silicate geochemical cycle and its effect on atmospheric carbondioxide over the past 100 million years, Am. J. Sci., 283, 641–683, 1983.

Boetius, A., Ravenschlag, K., Schubert, C. J., et al.: A marine mi-crobial consortium apparently mediating anaerobic oxidation ofmethane, Nature, 407(6804), 623–626, 2000.

Borowski, W. S.: A review of methane and gas hydrates in thedynamic, stratified system of the Blake Ridge region, offshoresoutheastern North America, Chem. Geol., 205(3–4), 311–346,2004.

Borowski, W. S., Paull, C. K., and Ussler, W.: Marine pore-watersulfate profiles indicate in situ methane flux from underlying gashydrate, Geology, 24(7), 655–658, 1996.

Borowski, W. S., Paull, C. K.. and Ussler, W.: Global and localvariations of interstitial sulfate gradients in deep-water, continen-tal margin sediments: Sensitivity to underlying methane and gashydrates, Mar. Geol., 159(1–4), 131–154, 1999.

Boudreau, B. P., Algar, C., Johnson, B. D., et al.: Bubble growthand rise in soft sediments, Geology, 33(6), 517–520, 2005.

Bowen, G. J., Beerling, D. J., Koch, P. L., et al.: A humid climatestate during the Palaeocene/Eocene thermal maximum, Nature,432(7016), 495–499, 2004.

Bralower, T. J., Thomas, D. J., Zachos, J. C., et al.: High-resolutionrecords of the late Paleocene thermal maximum and circum-Caribbean volcanism: Is there a causal link?, Geology, 25(11),963–966, 1997.

Brewer, P. G., Paull, C., Peltzer, E. T., et al.: Measure-ments of the fate of gas hydrates during transit throughthe ocean water column, Geophys. Res. Lett., 29(22), 2081,doi:10.1029/2002GL014727, 2002.

Brook, E. J., Harder, S., Severinghaus J., et al.: On the origin andtiming of rapid changes in atmospheric methane during the lastglacial period, Global Biogeochem. Cy., 14(2), 559–572, 2000.

Brooks, J. M., Bryant, W. R., Bernard, B. B., et al.: The nature ofgas hydrates on the Nigerian continental slope, in: Gas Hydrates:Challenges for the Future, edited by: Holder, G. D. and Bishnoi,P. R., 76–93, 2000.

Bryn, P., Berg, K., Forsberg, C. F., et al.: Explaining the StoreggaSlide, Mari. Petrol. Geol., 22(1–2), 11–19, 2005.

Bubier, J., Moore, T., Savage, K., et al.: A comparison of methaneflux in a boreal landscape between a dry and a wet year, GlobalBiogeochem. Cy., 19(1), GB1023, doi:10.1029/2004GB002351,2005.

Bubier, J. L., Moore, T. R., Bellisario, L., et al.: Ecological Controlson Methane Emissions from a Northern Peatland Complex in theZone of Discontinuous Permafrost, Manitoba, Canada, GlobalBiogeochem. Cy., 9(4), 455–470, 1995.

Buffett, B. and Archer, D. E.: Global inventory of methaneclathrate: Sensitivity to changes in environmental conditions,Earth Planet. Sci. Lett., 227, 185–199, 2004.

Buffett, B. A.: Clathrate hydrates, Annu. Rev. Earth Planet. Sci.,28, 477–507, 2000.

Buffett, B. A. and Zatsepina, O. Y.: Metastability of gas hydrate,Geophys. Res. Lett., 26(19), 2981–2984, 1999.

Buffett, B. A. and Zatsepina, O. Y.: Formation of gas hydrate fromdissolved gas in natural porous media, Mar. Geol., 164(1–2), 69–77, 2000.

Bunz, S. and Mienert, J.: Acoustic imaging of gas hydrate and freegas at the Storegga Slide, J. Geophys. Res.-Solid Earth, 109(B4),B04102, doi:10.1029/2003JB002863, 2004.

Camill, P.: Permafrost thaw accelerates in boreal peatlands duringlate-20th century climate warming, Clim. Change, 68(1–2), 135–152, 2005.

Cannariato, K. G. and Stott, L. D.: Evidence againstclathrate-derived methane release to Santa Barbara Basinsurface waters?, Geochem. Geophys. Geosyst., 5, Q05007,doi:10.1029/2003GC000600, 2004.

Chatti, I., Delahaye, A. Fournaison, L., et al.: Benefits and draw-backs of clathrate hydrates: a review of their areas of interest,



Energ. Convers. Manage., 46(9–10), 1333–1343, 2005.Circone, S., Stern, L. A., Kirby, S. H., et al.: Methane hydrate dis-

sociation rates at 0.1 MPa and temperatures above 272 K, in: GasHydrates: Challenges for the Future edited by: Holder, G. D. andBishnoi, P. R., 544–555, 2000.

Clark, J. F., Washburn, L., Hornafius, J. S., et al.: Dissolved hydro-carbon flux from natural marine seeps to the southern CaliforniaBight, J. Geophys. Res.-Oceans, 105(C5), 11 509–11 522, 2000.

Clennell, M. B., Henry, P., Hovland, M., et al.: Formation of nat-ural gas hydrates in marine sediments. Gas hydrate growth andstability conditioned by host sediment properties, in: Gas Hy-drates: Challenges for the Future, edited by: Holder, G. D. andBishnoi, P. R., 887–896, 2000.

Clennell, M. B., Hovland, M., Booth, J. S., et al.: Formation ofnatural gas hydrates in marine sediments 1. Conceptual modelof gas hydrate growth conditiond by host sediment properties, J.Geophys. Res. B, 104, 22 985–23 003, 1999.

Collett, T. S. and Kuuskraa, V. A.: Hydrates contain vast store ofworld gas resources, Oil Gas J., 96, 90–95, 1998.

D’Hondt, S., Jorgensen, B. B., Miller, D. J., et al.: Distributionsof microbial activities in deep subseafloor sediments, Science,306(5705), 2216–2221, 2004.

D’Hondt, S., Rutherford, S., and Spivack, A. J.: Metabolic activ-ity of subsurface life in deep-sea sediments, Science, 295(5562),2067–2070, 2002.

Dallenbach, A., Blunier, T., Fluckigeret, J., et al.: Changes in the at-mospheric CH4 gradient between Greenland and Antarctica dur-ing the Last Glacial and the transition to the Holocene, Geophys.Res. Lett., 27, 1005–1008, 2000.

Dallimore, S. R. and Collett, T. S.: Intrapermafrost Gas Hydratesfrom a Deep Core-Hole in the Mackenzie Delta, Northwest-Territories, Canada, Geology, 23(6), 527–530, 1995.

Dansgaard, W., White, J. W. C., and Johnson, S. J.: The abrupttermination of the Younger Dryas climate event, Nature, 339,532–534, 1989.

Davie, M. K. and Buffett, B. A.: A numerical model for the forma-tion of gas hydrate below the seafloor, J. Geophys. Res., 106(B1),497–514, 2001.

Deffeyes, K. S.: Hubbert’s Peak: The Impending World Oil Short-age, Princeton University Press, Princeton, NJ, 2001

Dickens, G. R.: The potential volume of oceanic methane hydrateswith variable external conditions, Org. Geochem., 32, 1179–1193, 2001.

Dickens, G. R., O’Heill, J. R., Rea, D. K., et al.: Dissociation ofoceanic methane hydrate as a cause of the carbon isotope excur-sion at the end of the Paleocene, Paleoceanography, 19, 965–971,1995.

Dickens, G. R., Paull, C. K., and Wallace, P.: Direct measurementof in situ methane quantities in a large gas-hydrate reservoir, Na-ture, 385(6615), 426–428, 1997.

Dimitrov, L. and Woodside, J.: Deep sea pockmark environments inthe eastern Mediterranean, Mar. Geol., 195(1–4), 263–276, 2003.

Dimitrov, L. I.: Mud volcanoes - the most important pathway fordegassing deeply buried sediments, Earth-Sci. Rev., 59(1–4), 49–76, 2002.

Driscoll, N. W., Weissel, J. K., and Goff, J. A.: Potential for large-scale submarine slope failure and tsunami generation along theUS mid-Atlantic coast, Geology, 28(5), 407–410, 2000.

Dugan, B. and Flemings, P. B.: Overpressure and fluid flow in the

New Jersey continental slope: Implications for slope failure andcold seeps, Science, 289, 288–291, 2000.

Egeberg, P. K. and Barth, T.: Contribution of dissolved organicspecies to the carbon and energy budgets of hydrate bearing deepsea sediments (Ocean Drilling Program Site 997 Blake Ridge),Chem. Geol., 149(1–2), 25–35, 1998.

Fehn, U., Snyder, G., and Egeberg, P. K.: Dating of pore waters withI-129: Relevance for the origin of marine gas hydrates, Science,289(5488), 2332–2335, 2000.

Flemings, B. P., Liu, X., and Winters, W. J.: Critical pressure andmultiphase flow in Blake Ridge gas hydrates, Geology, 31, 1057–1060, 2003.

Flemings, P. B., Stump, B. B., Finkbeiner, T., et al.: Flow focusingin overpressured sandstones: Theory, observations, and applica-tions, Am. J. Sci., 302, 827–855, 2002.

Gardiner, B. S., Boudreau, B. P., and Johnson, B. D.: Growth ofdisk-shaped bubbles in sediments, Geochim. Cosmochim. Ac.,67(8), 1485–1494, 2003.

Gavrilov, A. V., Romanovskii, X. N., Romanovsky, V. E., et al.:Reconstruction of ice complex remnants on the eastern SiberianArctic Shelf, Permafrost Periglac., 14(2), 187–198, 2003.

Gorman, A. R., Holbrook, W. S., Hornbach, M. J., et al.: Migrationof methane gas through the hydrate stability zone in a low-fluxhydrate province, Geology, 30(4), 327–330, 2002.

Gornitz, V. and Fung, I.: Potential distribution of methane hydratein the world’s oceans, Global Biogeochem. Cy., 8, 335–347,1994.

Grace, J. D. and Hart, G. F.: Giant Gas-Fields of Northern WestSiberia, Aapg Bull., 70(7), 830–852, 1986.

Grant, N. J. and Whiticar, M. J.: Stable carbon isotopic evidencefor methane oxidation in plumes above Hydrate Ridge, Cas-cadia Oregon Margin, Global Biogeochem. Cy., 16(4), 1124,doi:10.1029/2001GB001851, 2002.

Grauls, D.: Gas hydrates: importance and applications in petroleumexploration, Mar. Petrol. Geol., 18(4), 519–523, 2001.

Hampton, M. A., Lee, H. J., and Locat, J.: Submarine landslides,Rev. Geophys., 34(1), 33–59, 1996.

Hansen, J., Sato, M., Ruedy, R., et al.: Efficacy of cli-mate forcings, J. Geophys. Res.-Atmos., 110(D18), D18104,doi:10.1029/2005JD005776, 2005.

Harvey, L. D. D. and Huang, Z.: Evaluation of the potential impactof methane clathrate destabilization on future global warming, J.Geophys. Res., 100, 2905–2926, 1995.

Heeschen, K. U., Collier, R. W., de Angelis, M. A., et al.: Methanesources, distributions, and fluxes from cold vent sites at Hy-drate Ridge, Cascadia Margin, Global Biogeochem. Cy., 19(2),GB2016, doi:10.1029/2004GB002266, 2005.

Heeschen, K. U., Trehu, A. M., Collier, R. W., et al.: Distributionand height of methane bubble plumes on the Cascadia Margincharacterized by acoustic imaging, Geophys. Res. Lett., 30(12),1643, doi:10.1029/2003GL016974, 2003.

Herz, O. F.: Frozen mammoths in Siberia, Ann. Rep. SmithsonianInst., 1904, 611–625, 1904.

Hesselbo, S. P., Grocke, D. R., Jenkyns, H. C., et al.: Massive dis-sociation of gas hydrate during a Jurassic oceanic anoxic event,Nature, 406(6794), 392–395, 2000.

Hill, J. C., Driscoll, N. W., Weissel, J. K., et al.: Large-scale elongated gas blowouts along the US Atlantic mar-gin, J. Geophys. Res.-Solid Earth, 109(B9), B09101,



doi:10.1029/2004JB002969, 2004.Hinrichs, K. U., Hmelo L. R., and Sylva, S. P.: Molecular fossil

record of elevated methane levels in late pleistocene coastal wa-ters, Science, 299(5610), 1214–1217, 2003.

Hoehler, T. M., Albert, D. B., Alperin, M. J., et al.: Acetogene-sis from CO2 in an anoxic marine sediment, Limnol. Oceanogr.,44(3), 662–667, 1999.

Hoehler, T. M., Alperin, M. J., Albert, D. B., et al.: Thermody-namic control on hydrogen concentrations in anoxic sediments,Geochim. Cosmochim. Ac., 62(10), 1745–1756, 1998.

Hornbach, M. J., Saffer, D. M., and Holbrook, W. S.: Criticallypressured free-gas reservoirs below gas-hydrate provinces, Na-ture, 427(6970), 142–144, 2004.

Hovland, M., Gallagher, J. W., and Clennell, M. B., et al.: Gas hy-drate and free gas volumes in marine sediments: Example fromthe Niger Delta front, Mar. Petrol. Geol., 14(3), 245–255, 1997.

Hovland, M. and Judd, A.: Seabed pockmarks and seepages: im-pact on geology, biology, and the marine environment, Graham& Trotman, London, 1988

Hovland, M., Svensen, H., Forsberg, C. F., et al.: Complex pock-marks with carbonate-ridges off mid-Norway: Products of sedi-ment degassing, Mar. Geol., 218(1–4), 191–206, 2005.

Hubberten, H. W. and Romanovskii, N. N.: Terrestrial and offshorepermafrost evolution of the Laptev sea region during the lastPleistocene-Holocene glacial-eustatic cycle, in: Permafrost re-sponse on economic develoopment, environmental security andnatural resources, edited by: Paepa, R. and Melnikov, V., Klewer,43–60, 2001.

Huber, M., Sloan, L., and Shellito, C.: Early Paleogene oceans andclimate: A fully coupled modeling approach using the NationalCenter for Atmospheric Research Community Climate SystemModel, in: Causes and consequences of globally warm climatesin the Early Paleogene, edited by: Wing, S. L., Gingerich, P. D.,Schmitz, B., and Thomas, E., Geological Society of America,25–47, 2002.

IPCC, 2007 Climate Change 2007: The Physical Science Basis,edited by: Solomon, S., Qin, D., Manning, M., Marquis, M.,Averyt, K., Tignor, M. M. B., Miller, H. L., and Chen, Z., Cam-bridge University Press.

Jenkyns, H. C.: Evidence for rapid climate change in the Mesozoic-Palaeogene greenhouse world, Philosophical Transactions of theRoyal Society of London Series a-Mathematical Physical andEngineering Sciences, 361(1810), 1885–1916, 2003.

Kennett, J. P., Cannariato, K. G., Hendy, I. L., et al.: Carbon iso-topic evidence for methane hydrate instability during quaternaryinterstadials, Science, 288(5463), 128–133, 2000.

Kennett, J. P., Cannariato, K. G., Hendy, I. L., et al.: Methane hy-drates in Quaternary climate change: The clathrate gun hypothe-sis, AGU, 2003

Kennett, J. P. and Stott, L. D.: Abrupt deep sea warming, paleo-ceanographic changes and benthic extinctions at the end of thePaleocene, Nature, 353, 319–322, 1991.

Kent, D. V., Cramer, B. S., Lanci, L., et al.: A case for a cometimpact trigger for the Paleocene/Eocene thermal maximum andcarbon isotope excursion, Earth Planet. Sci. Lett., 211(1–2), 13–26, 2003.

Kerr, R. A.: Energy – Gas hydrate resource: Smaller but sooner,Science, 303(5660), 946–947, 2004.

Klauda, J. B. and Sandler, S. I.: Global distribution of methane hy-

drate in ocean sediment, Energy & Fuels, 19(2), 459–470, 2005.Koch, P. L., Zachos, J. C., and Gingerich, P.D.: Coupled isotopic

change in marine and continental carbon reservoirs near the Pa-leocene/Eocene boundary, Nature, 358, 319–322, 1992.

Krason, J.: Messoyakh gas field (W. Siberia) – A model for devel-opment of the methane hydrate deposits of Mackenzie Delta, in:Gas Hydrates: Challenges for the Future, edited by: Holder, G.D. and Bishnoi, P. R., 173–188, 2000.

Kump, L. R. and Arthur, M. A.: Interpreting carbon-isotope excur-sions: carbonates and organic matter, Chem. Geol., 161, 181–198, 1999.

Kvenvolden, K. A.: Methane Hydrate – a Major Reservoir of Car-bon in the Shallow Geosphere, Chem. Geol., 71(1–3), 41–51,1988.

Kvenvolden, K. A.: Potential effects of gas hydrate on human wel-fare, Proc. Natl. Acad. Sci. USA, 96, 3420–3426, 1999.

Larson, R. L.: Geological consequences of superplumes, Geology,19, 963–966, 1991.

Leifer, I., Clark, J. F., and Chen, R. F.: Modifications of the local en-vironment by natural marine hydrocarbon seeps, Geophys. Res.Lett., 27(22), 3711–3714, 2000.

Leifer, I. and MacDonald, I.: Dynamics of the gas flux from shallowgas hydrate deposits: interaction between oily hydrate bubblesand the oceanic environment, Earth Planet. Sci. Lett., 210(3–4),411–424, 2003.

Levitus, S., Conkright, M. E., Reid, J. L., et al.: Distributionof nitrate, phosphate, and silicate in the world’s oceans, Prog.Oceanogr., 31, 245–273, 1993.

Levitus, S., Antonov, J., and Boyer, T.: Warming of the worldocean, 1955–2003, Geophys. Res. Lett., 32(2),L02604, L02604,doi:10.1029/2004GL021592, 2005.

Liblik, L. K., Moore, T. R., Bubier, J. L.: Methane emissions fromwetlands in the zone of discontinuous permafrost: Fort Simpson,Northwest Territories, Canada, Global Biogeochem. Cy., 11(4),485–494, 1997.

Lorenson, T. D.: Microscopic character of marine sediment contain-ing disseminated gas hydrate – Examples from the slake ridgeand the middle America Trench, in: Gas Hydrates: Challengesfor the Future, edited by: Holder, G. D. and Bishnoi, P. R., 189–194, 2000.

Lourens, L. J., Sluijs, A., Kroon, D., et al.: Astronomical pacing oflate Palaeocene to early Eocene global warming events, Nature,435(7045), 1083–1087, 2005.

Luff, R., Greinert, J., Wallmann, K., et al.: Simulation of long-termfeedbacks from authigenic carbonate crust formation at cold ventsites, Chem. Geol., 216(1–2), 157–174, 2005.

Luyendyk, B., Kennett, J., and Clark, J. F.: Hypothesis for in-creased atmospheric methane input from hydrocarbon seeps onexposed continental shelves during glacial low sea level, Mar.Petrol. Geol., 22(4), 591–596, 2005.

Luyten, J. R., Pedlosky, J., and Stommel, H.: The ventilated ther-mocline, J. Phys. Ocean., 13, 292–309, 1983.

MacDonald, G.: Role of methane clathrates in past and future cli-mates, Clim. Change, 16, 247–281, 1990.

MacDonald, I. R., Bohrmann, G., Escobar, E., et al.: Asphalt vol-canism and chemosynthetic life in the Campeche Knolls, Gulf ofMexico, Science, 304(5673), 999–1002, 2004.

Macdonald, I. R., Guinasso, N. L., Sassen, R., et al.: Gas Hy-drate That Breaches the Sea-Floor on the Continental-Slope of



the Gulf-of-Mexico, Geology, 22(8), 699–702, 1994.MacDonald, I. R., Leifer, I., Sassen, R., et al.: Transfer of hydro-

carbons from natural seeps to the water column and atmosphere,Geofluids, 2(2), 95–107, 2002.

Martens, C. S., Albert, D. B., and Alperin, M. J.: Biogeochemicalprocesses controlling methane in gassy coastal sediments – Part1. A model coupling organic matter flux to gas production, ox-idation and transport, Cont. Shelf Res., 18(14–15), 1741–1770,1998.

Martin, P., Archer, D., and Lea, D.: Role of deep sea temperaturesin the carbon cycle during the last glacial, Paleoceanography,PA2015, doi:10.1029/2003PA000914, 2005.

Maslin, M., Owen, M., Day, S., et al.: Linking continental-slopefailures and climate change: Testing the clathrate gun hypothesis,Geology, 32(1), 53–56, 2004.

Mienert, J., Andreassen, K., Posewang, J., et al.: Changes of thehydrate stability zone of the Norwegian margin from glacial tointerglacial times, in: Gas Hydrates: Challenges for the Future,edited by: Holder, G. D. and Bishnoi, P. R., 200—210, 2000.

Mienert, J., Vanneste, M., Bunz, S., et al.: Ocean warming and gashydrate stability on the mid-Norwegian margin at the StoreggaSlide, Mar. Petrol. Geol., 22(1–2), 233–244, 2005.

Milkov, A. V.: Worldwide distribution of submarine mud volcanoesand associated gas hydrates, Mar. Geol., 167(1–2), 29–42, 2000.

Milkov, A. V.: Global estimates of hydrate-bound gas in marinesediments: how much is really out there?, Earth-Sci. Rev., 66(3–4), 183–197, 2004.

Milkov, A. V.: Molecular and stable isotope compositions of naturalgas hydrates: A revised global dataset and basic interpretationsin the context of geological settings, Org. Geochem., 36(5), 681–702, 2005.

Milkov, A. V. and Sassen, R.: Thickness of the gas hydrate stabilityzone, Gulf of Mexico continental slope, Mar. Petrol. Geol., 17(9),981–991, 2000.

Milkov, A. V. and Sassen, R.: Estimate of gas hydrate resource,northwestern Gulf of Mexico continental slope, Mar. Geol.,179(1–2), 71–83, 2001.

Milkov, A. V. and Sassen, R.: Economic geology of offshore gashydrate accumulations and provinces, Mar. Petrol. Geol., 19(1),1–11, 2002.

Milkov, A. V. and Sassen, R.: Two-dimensional modeling of gashydrate decomposition in the northwestern Gulf of Mexico: sig-nificance to global change assessment, Global Planet. Change,36(1–2), 31–46, 2003.

Milkov, A. V., Vogt, P. R., Crane, K., et al.: Geological, geochemi-cal, and microbial processes at the hydrate-bearing Hakon Mosbymud volcano: a review, Chem. Geol., 205(3–4), 347–366, 2004.

Nelson, F. E., Anisimov, O. A., and Shiklomanov, N. I.: Climatechange and hazard zonation in the circum-Arctic permafrost re-gions, Nat. Hazards, 26(3), 203–225, 2002.

Nisbet, E. G.: Have sudden large releases of methane from geo-logical reservoirs occurred since the Last Glacial Maximum, andcould such releases occur again?, Philosophical Transactions ofthe Royal Society of London Series a-Mathematical Physical andEngineering Sciences, 360(1793), 581–607, 2002.

Nisbet, E. G. and Piper, D. J. W.: Giant submarine landslides, Na-ture, 392(6674), 329–330, 1998.

Nouze, H., Henry, P., Noble, M., et al.: Large gas hydrate ac-cumulations on the eastern Nankai Trough inferred from new

high-resolution 2-D seismic data, Geophys. Res. Lett., 31(13),L13308, doi:10.1029/2004GL019848, 2004.

Orcutt, B. N., Boetius, A., Lugo, S. K., et al.: Life at the edge ofmethane ice: microbial cycling of carbon and sulfur in Gulf ofMexico gas hydrates, Chem. Geol., 205(3–4), 239–251, 2004.

Pagani, M., Caldeira, K., Archer, D., and Zachos, J.: An ancientcarbon mystery, Science, 314, 1556–1557, 2006.

Parkes, R. J., Cragg, B. A., and Wellsbury, P.: Recent studies onbacterial populations and processes in subseafloor sediments: Areview, Hydrogeol. J., 8(1), 11–28, 2000.

Paul, A. Z.: The effect of benthic biological processes on the CO2carbonate system, in: The Fate of Fossil Fuel CO2 in the Oceans,edited by: A. M. A. Andersen N.R., Plenum Press, 345, 1978.

Paull, C. K., Brewer, P. G., Ussler, W., et al.: An experiment demon-strating that marine slumping is a mechanism to transfer methanefrom seafloor gas-hydrate deposits into the upper ocean and at-mosphere, Geo-Mar. Lett., 22(4), 198–203, 2003.

Paull, C. K., Lorenson, T. D., Dickens, G., et al.: Comparisons ofin situ and core gas measurements in ODP Leg 164 bore holes,in: Gas Hydrates: Challenges for the Future, edited by: Holder,G. D. and Bishnoi, P. R., 23–31, 2000.

Pavlov, A. A., Kasting, J. F., Brown, L. L., et al.: Greenhouse warm-ing by CH4 in the atmosphere of early Earth, J. Geophys. Res.-Planets, 105(E5), 11 981–11 990, 2000.

Payette, S., Delwaide, A., Caccianiga, M., et al.: Accelerated thaw-ing of subarctic peatland permafrost over the last 50 years, Geo-phys. Res. Lett., 31, L18208, doi:10.1029/2004GL020358, 2004.

Pecher, I. A., Kukowski, N., Huebscher, C., et al.: The link be-tween bottom-simulating reflections and methane flux into thegas hydrate stability zone – new evidence from Lima Basin, PeruMargin, Earth Planet. Sci. Lett., 185(3–4), 343–354, 2001.

Rehder, G., Brewer, P. W., Peltzer, E. T., et al.: Enhanced lifetimeof methane bubble streams within the deep ocean, Geophys. Res.Lett., 29(15), 1731, doi:10.1029/2001GL013966, 2002.

Rehder, G., Keir, R. S., Suess, E., et al.: Methane in the northern At-lantic controlled by microbial oxidation and atmospheric history,Geophys. Res. Lett., 26(5), 587–590, 1999.

Rehder, G., Kirby, S. H., Durham, W. B., et al.: Dissolution ratesof pure methane hydrate and carbon-dioxide hydrate in under-saturated seawater at 1000-m depth, Geochim. Cosmochim. Ac.,68(2), 285–292, 2004.

Riedel, M., Spence, G. D., Chapman, N. R., et al.: Seismic inves-tigations of a vent field associated with gas hydrates, offshoreVancouver Island, J. Geophys. Res.-Solid Earth, 107(B9), 2200,doi:10.1029/2001JB000269, 2002.

Rivkina, E., Laurinavichius, K., McGrath, J., et al.: Microbial lifein permafrost, Adv. Space Res., 33, 1215–1221, 2004.

Rivkina, E. M., Friedmann, E. I., McKay, C. P., et al.: Metabolicactivity of permafrost bacteria below the freezing point, Appl.Environ. Microbiol., 66(8), 3230–3233, 2000.

Roehl, U., Bralower, T.J., Norris, G., and Wefer, G.: A new chronol-ogy for the late Paleocene thermal maximum and its environmen-tal implications, Geology, 28, 927–930, 2000.

Rogner, H.-H.: An assessment of world hydrocarbon resources,Ann. Rev. Energy Environ., 22, 217–262, 1997.

Romankevich, E. A.: Geochemistry of Organic Matter in the Ocean,Springer, New York, 1984

Rothman, L. S.: Atmospheric propagation of radiation – microwavethrough visible, Onde Electrique, 72(3), 39–41, 1992.



Rothwell, R. G., Reeder, M. S., Anastasakis, G., et al.: Low sea-level stand emplacement of megaturbidites in the western andeastern Mediterranean Sea, Sediment. Geol., 135(1–4), 75–88,2000.

Sager, W. W., Lee, C. S., Macdonald, I. R., et al.: High-frequencynear-bottom acoustic reflection signatures of hydrocarbon seepson the Northern Gulf of Mexico continental slope, Geo-Mar.Lett., 18(4), 267–276, 1999.

Sassen, R., Losh, S. L., Cathles, L., et al.: Massive vein-filling gashydrate: relation to ongoing gas migration from the deep subsur-face in the Gulf of Mexico, Mar. Petrol. Geol., 18(5), 551–560,2001.

Sassen, R. and MacDonald, I. R.: Evidence of structure H hydrate,Gulf of Mexico continental slope, Org. Geochem., 22, 1029–1032, 1994.

Sassen, R., Milkov, A. V., Roberts, H. H., et al.: Geochemical evi-dence of rapid hydrocarbon venting from a seafloor-piercing muddiapir, Gulf of Mexico continental shelf, Mar. Geol., 198(3–4),319–329, 2003.

Sassen, R., Sweet, S. T., Milkov, A. V., et al.: Thermogenic ventgas and gas hydrate in the Gulf of Mexico slope: Is gas hydratedecomposition significant?, Geology, 29(2), 107–110, 2001.

Sazonova, T. S., Romanovsky, V. E., Walsh, J. E., et al.: Per-mafrost dynamics in the 20th and 21st centuries along the EastSiberian transect, J. Geophys. Res.-Atmos., 109(D1), D01108,doi:10.1029/2003JD003680, 2004.

Schmidt, G. A. and Shindell, D. T.: Atmospheric composition,radiative forcing, and climate change as a consequence of amassive methane release from gas hydrates, Paleoceanography,18(1), 1004, doi:10.1029/2002PA000757, 2003.

Schmitz, B., Peucker-Ehrenbrink, B., Heilmann-Clausen, C., etal.: Basaltic explosive volcanism, but no comet impact, at thePaleocene-Eocene boundary: high-resolution chemical and iso-topic records from Egypt, Spain and Denmark, Earth Planet. Sci.Lett., 225(1–2), 1–17, 2004.

Shakhova, N., Semiletov, I., and Panteleev, G.: The distributionof methane on the Siberian Arctic shelves: Implications forthe marine methane cycle, Geophys. Res. Lett., 32(9), L09601,doi:10.1029/2005GL022751, 2005.

Sloan, E. D.: Clathrate hydrates of natural gas, Marcel Dekker, NewYork, 1998.

Sluijs, A., Schouten, S., Pagani, M., et al.: Subtropical Arctic Oceantemperatures during the Palaeocene/Eocene thermal maximum,Nature, 441(7093), 610–613, 2006.

Smith, D. E., Shi, S., Cullingford, R. A., et al.: The holocenestoregga slide tsunami in the United Kingdom, Quaternary Sci.Rev., 23(23–24), 2291–2321, 2004.

Smith, S. L., Burgess, M. M., Riseborough, D., et al.: Recent trendsfrom Canadian Permafrost Thermal Monitoring Network Sites,Permafrost Periglac., 16(1), 19–30, 2005.

Solheim, A., Berg, K., Forsberg, C. F., et al.: The Storegga Slidecomplex: repetitive large scale sliding with similar cause anddevelopment, Mar. Petrol. Geol., 22(1–2), 97–107, 2005.

Sowers, T.: Late quaternary atmospheric CH4 isotope record sug-gests marine clathrates are stable, Science, 311(5762), 838–840,2006.

Stendel, M. and Christensen, J. H.: Impact of global warming onpermafrost conditions in a coupled GCM, Geophys. Res. Lett.,29(13), 1632, doi:10.1029/2001GL014345, 2002.

Stockstad, E.: Defrosting the carbon freezer of the North, Science,304, 1618–1620, 2004.

Suess, E., Torres, M. E., Bohrmann, G., et al.: Gas hydrate desta-bilization: enhanced dewatering, benthic material turnover andlarge methane plumes at the Cascadia convergent margin, EarthPlanet. Sci. Lett., 170(1–2), 1–15, 1999.

Sultan, N., Cochonat, P., Foucher, J. P., et al.: Effect of gas hydratesmelting on seafloor slope instability, Mar. Geol., 213(1–4), 379–401, 2004.

Svensen, H., Planke, S., Malthe-Sorenssen, A., et al.: Release ofmethane from a volcanic basin as a mechanism for initial Eoceneglobal warming, Nature, 429(6991), 542–545, 2004.

Taylor, M. H., Dillon, W. P., and Pecher, I. A.: Trapping and mi-gration of methane associated with the gas hydrate stability zoneat the Blake Ridge Diapir: new insights from seismic data, Mar.Geol., 164(1–2), 79–89, 2000.

Teichert, B. M. A., Eisenhauer, A., Bohrmann, G., et al.: U/Th sys-tematics and ages of authigenic carbonates from Hydrate Ridge,Cascadia Margin: Recorders of fluid flow variations, Geochim.Cosmochim. Ac., 67(20), 3845–3857, 2003.

Teichert, B. M. A., Gussone, N., Eisenhauer, A., et al.: Clathrites:Archives of near-seafloor pore-fluid evolution (delta Ca-44/40,delta C-13, delta O-18) in gas hydrate environments, Geology,33(3), 213–216, 2005.

Thomas, D. J., Zachos, J. C., Bralower, T. J., et al.: Warming thefuel for the fire: Evidence for the thermal dissociation of methanehydrate during the Paleocene-Eocene thermal maximum, Geol-ogy, 30(12), 1067–1070, 2002.

Thorpe, R. B., Law, K. S., Bekki, S., et al.: Is methane-drivendeglaciation consistent with the ice core record?, J. Geophys.Res.-Atmos., 101(D22), 28 627–28 635, 1996.

Torres, M. E., McManus, J., Hammond, D. E., et al.: Fluid andchemical fluxes in and out of sediments hosting methane hydratedeposits on Hydrate Ridge, OR, I: Hydrological provinces, EarthPlanet. Sci. Lett., 201(3–4), 525–540, 2002.

Torres, M. E., Wallmann, K., Trehu, A. M., et al.: Gas hydrategrowth, methane transport, and chloride enrichment at the south-ern summit of Hydrate Ridge, Cascadia margin off Oregon, EarthPlanet. Sci. Lett., 226(1–2), 225–241, 2004.

Trehu, A. M., Flemings, P. B., Bangs, N. L., et al.: Feeding methanevents and gas hydrate deposits at south Hydrate Ridge, Geophys.Res. Lett., 31(23), L23310, doi:10.1029/2004GL021286, 2004.

Trehu, A. M., Long, P. E., Torres, M. E., et al.: Three-dimensionaldistribution of gas hydrate beneath southern Hydrate Ridge: con-straints from ODP Leg 204, Earth Planet. Sci. Lett., 222(3–4),845–862, 2004.

Uchida, M., Shibata, Y., Ohkushi, K, et al.: Episodic methane re-lease events from Last Glacial marginal sediments in the west-ern North Pacific, Geochem. Geophys. Geosyst., 5, Q08005,doi:10.1029/2004GC000699, 2004.

Valentine, D. L., Blanton, D. C., Reeburgh, W. S., et al.: Water col-umn methane oxidation adjacent to an area of active hydrate dis-sociation, Eel River Basin, Geochim. Cosmochim. Ac., 65(16),2633–2640, 2001.

Vogt, P. R. and Jung, W. Y.: Holocene mass wasting on uppernon-Polar continental slopes – due to post-Glacial ocean warm-ing and hydrate dissociation?, Geophys. Res. Let., 29(9), 1341,doi:10.1029/2001GL013488, 2002.

Wang, Y. and Jacob, D. J.: Anthropogenic forcing on tropospheric



ozone and OH since preindustrial times, J. Geophys. Res., 103,31 123–31 135, 1998.

Warzinski, R. and Holder, G.: Gas clathrate hydrates, Energy &Fuels, 12(2), 189–190, 1998.

Washburn, L., Clark, J. F., and Kyriakidis, P.: The spatial scales,distribution, and intensity of natural marine hydrocarbon seepsnear Coal Oil Point, California, Mar. Petrol. Geol.y, 22(4), 569–578, 2005.

Weinberger, J. L., Brown, K. M., and Long, P. E.: Painting a pictureof gas hydrate distribution with thermal images, Geophys. Res.Lett., 32(4), L04609, doi:10.1029/2004GL021437, 2005.

Wellsbury, P., Goodman, K., Barth, T., et al.: Deep marine bio-sphere fuelled by increasing organic matter availability duringburial and heating, Nature, 388(6642), 573–576, 1997.

Wellsbury, P., Mather, I., and Parkes, R. J.: Geomicrobiology ofdeep, low organic carbon sediments in the Woodlark Basin, Pa-cific Ocean, Fems Microbiology Ecology, 42(1), 59–70, 2002.

Whiticar, M. J. and Faber, E.: Methane Oxidation in Sedimentand Water Column Environments - Isotope Evidence, Org.Geochem., 10(4–6), 759–768, 1986.

Winters, W. J., Pecher, I. A., Waite, W. F.: Physical propertiesand rock physics models of sediment containing natural andlaboratory-formed methane gas hydrate, American Mineralogist,89(8–9), 1221–1227, 2004.

Wood, W. T., Gettrust, J. F., Chapman, N. R., et al.: Decreased sta-bility of methane hydrates in marine sediments owing to phase-boundary roughness, Nature, 420(6916), 656–660, 2002.

Xu, W. Y., Lowell, R. P., and Peltzer, E. T.: Effect of seafloortemperature and pressure variations on methane flux from a gashydrate layer: Comparison between current and late Paleoceneclimate conditions, J. Geophys. Res.-Solid Earth, 106(B11),26 413–26 423, 2001.

Zachos, J. C., Pagani, M., Sloan, L., et al.: Trends, rhythms, andaberrations in global climate 65 Ma to Present, Science, 292,686–693, 2001.

Zachos, J. C., Rohl, U., Schellenberg, S. A., et al.: Rapid acidifi-cation of the ocean during the Paleocene-Eocene thermal maxi-mum, Science, 308(5728), 1611–1615, 2005.

Zachos, J. C., Wara, M. W., Bohaty, S., et al.: A transient risein tropical sea surface temperature during the Paleocene-EoceneThermal Maximum, Science, 302(5650), 1551–1554, 2003.

Zeebe, R. E. and Westbroek, P.: A simple model for the CaCO3saturation state of the ocean: The “Strangelove”, the “Neritan”,and the “Cretan” Ocean, Geochem. Geophys. Geosyst., 4, 1104,doi:10.1029/2003GC000538, 2003.

Zillmer, M., Flueh, E. R., Petersen, J., et al.: Seismic investigationof a bottom simulating reflector and quantification of gas hydratein the Black Sea, Geophys. J. Int., 161(3), 662–678, 2005a.

Zillmer, M., Reston, T., Leythaeuser, T., et al.: Imaging andquantification of gas hydrate and free gas at the Storeggaslide offshore Norway, Geophys. Res. Lett., 32(4), L04308,doi:10.1029/2004GL021535, 2005b.

Zuhlsdorff, L. and Spiess, V.: Three-dimensional seismic character-ization of a venting site reveals compelling indications of naturalhydraulic fracturing, Geology, 32(2), 101–104, 2004.

Zuhlsdorff, L., Spiess, V., Hubscher, C., et al.: Implications for fo-cused fluid transport at the northern Cascadia accretionary prismfrom a correlation between BSR occurrence and near-sea-floorreflectivity anomalies imaged in a multi-frequency seismic dataset, Int. J. Earth Sci., 88(4), 655–667, 2000.


Documents

Methane hydrate stability and anthropogenic climate changegeosci.uchicago.edu/~archer/reprints/archer.2007.hydrate_rev.pdf · D. Archer: Methane hydrate stability and anthropogenic